Conference 7.286::digital

    *  What an awesome technical achievement!  Something to be proud of.
    
    *  What a WELL-DONE "Kudos" section, too.  People to be proud of!
    
    

Thanks for posting that. It's so nice to see the good things. It's also nice
to see people's work held up as "best in the world" and to see the time taken
to give credit where credit is due.

I hope that when the time comes, we can wallpaper the world with that report.

Mickey.

	This is fantastic news. Now... not to rain on anyones parade
	but, do any of the major business application suites run on
	this software? For example, ASK ManMan, SAP R/3, etc.

	I know that R/3 doesn't because we have purchased it for use within
	DEC. It does not support Rdb (it uses ORACLE, SYBASE and Informix)
	and uses its own TP monitor since ACMS and the like are not
	operable across hardware platforms.

	I think we should push this announcement hard to get customers
	of the application software vendors to demand that they port
	their software to the world's fastest machines.

    >>> I think we should push this announcement hard to get customers
    >>> of the application software vendors to demand that they port
    >>> their software to the world's fastest machines.

        Absolutely!  Let's make them chase us for a change.  
        Marketing is a poweful thing.  Just ask IBM, Anheuser-Bush, 
        and Microsoft.  Imagine what it can do when it has some
        really great products behind it.

    Terry

	I hope I've got you worried for a lot more than a second.

	We have had top notch products in a number of areas for 
	a long time that just never seem to make any waves.

	We will have the same thing here unless we can get the
	BUSINESS APPLICATION SOFTWARE that people use/demand
	supported on our platform.

	I've written client server stuff with ACMS and DesktopACMS
	and it works well, call it from Visual Basic etc. But, where's
	the market? Does ACMS run on HP; does Rdb run on IBM; etc.
	If our middleware only runs on Alpha then what are the odds
	that it will get built into 3rd party application suites?
	And it is those 3rd party applications that drive hardware
	decisions.


	Price performance is only a window of opportunity. HP et al
	will catch up eventually. We have to push brand recognition
	to the END USER to get pressure on the software vendors whose
	cooperation will make or break us. Imagine the conversation
	as someone from a large corporation tries to buy software and
	says "Oh, by the way, how many TPS can you get on the Alpha?"

	Wouldn't it be nice if there was an answer other than, "Oh,
	we don't run on DEC hardware, nobody asks for it."

	We have to market our product so that we are asked for BY NAME.
	The Intel Inside campaign is a good example. We need to generate
	a push from the end user to generate a pull on us from the 
	software vendors.

    
    Look at header of the base note "confidential , internal use only ".

    So, it has good chances to stay this way.

    Also, how relevant is it for the bulk of the market we target ?

    How is it relevant to the bulk of the market ?

    I don't know how many similar announces have we seen without any
    real realisation of the possible business potential.

	You ARE the base note.. What topic were you replying too?

						mike

I moved the previous two notes here after a new base note was erroneously
created.

Bob - Co-moderator DIGITAL

Hats off to the people who accomplished this, and the long line of supporting
members, including those who built the products that made it possible!

After having once worked on a project that attempted to create a custom
database and distributed transaction processing engine (because "RDB doesn't 
do what we need" and "the x industry doesn't like clusters"), this confirms
my belief that we should let the DB folks and the systems folks do their jobs,
'cause they know what they're doing. There's plenty of application
engineering work to do without trying to reinvent every wheel and screw
on our own.

A suggestion: publish a detailed description of the test as a case study in 
design, configuration, and tuning of a high-end TP system. How many customers
could benefit from the findings of the folks mentioned in the basenote? Advice
like that only comes along once in a lifetime; bottle it and sell it! That 
also makes a nice way of getting the word out. Here's what we did, and here's
how you can do it for your system. This would probably make a nice book.

    Echoing some of the sentiments of .3 and .5:
    
    I applaud the engineering efforts of all involved in reaching this
    important benchmark milestone. I have worked with several of the
    individuals involved over an 18 month period and have the greatest respect 
    for each of them.
    
    However, I have reservations about the business significance of this
    accomplishment. At a time when OpenVMS is waning and yielding to the
    UNIX-based systems in the marketplace, one has to ask, "so what?". I
    would be happier if we were publishing industry-leadership performance
    or price/performance results on Alpha/OSF or, better yet, generic UNIX
    systems. To date, we have promised, but not delivered, Rdb on
    Alpha/OSF and Alpha/NT only. What about the rest of the non-VMS world?

    Well as the person who posted the original here are some answers to
    those questions. 
    
    Re .3 (what business applications) There are many,many business
    applications that run on Rdb and the one you specifically mention
    (ManMan) runs on DBMS which is Rdb's sister product and would be able
    to perform as well as Rdb if it had been benchmarked in this particular
    way. SAP R/3 is as  you say but that might change in the future. Their
    main argument against doing this is that Rdb is not portable - well Rdb
    goes on OSF soon and NT is planned and others might follow.
    
    Re Wlodek. Considering this benchmark is the industry standard TPC-A
    benchmark which all the hardware vendors and software db vendors use to
    show how good their products are then this is pretty meaningful.
    
    Furthermore consider that the database market plus associated services
    is a huge chunk of business and Rdb is already one of the most
    profitable pieces of software that we sell. This will translate into
    dollars. 
    
    And the really good news is that I now understand that this will be
    announced publically shortly and so that this can be told to customers.
    
    Regards,
    
    Phil

	RE: .12
	
	It would be better if ManMan actually ran on Rdb don't you think?
	
	OSF and NT are not viable platforms either.  OSF is probably
	seen as proprietary to DEC  and NT is not yet a player.

	The db market is large but Rdb is an insignificant player there.
	The fact that the benchmark was obtained on OpenVMS is strike
	three.

	We can't yet do SMP on OSF not to mention clusters. 

	Again, it's a great technical achievement that I'm afraid
	won't get us very far.  I'm afraid that the time is past
	when a company could be vertically integrated and successful
	in the computer industry.

	Either you make software that runs on all the major platforms,
	or you make platforms that run  all the major software. Or
	you fill the niches.

	Trying to be a major player in both only causes trouble now.

    >    It would be better if ManMan actually ran on Rdb don't you think?
    
    I believe that Ask/Ingres has tried porting to a Relational database
    with poor results. If they could run on a relational db, I would think
    that they would want to port to their own db (Ingres).
    
    As to porting to non-digital platforms (hp, sun, rs6000); these are
    management issues.
    
    michael

>	We can't yet do SMP on OSF not to mention clusters. 

Weren't OSF clusters announced today under the name AdvantageCluster?

    
    
    Just a couple thoughts I had after reading the replies:
    
    - is there still a market for such performance?  Yea, I think so...for
    example MCI tracks all their 800 number calls with an Rdb database of
    about 100 GIG with txn rates over 1000/second.  No 'open' system can
    approach the performance and availability required for this
    system...YET.  They will of course catch up...
    
    - should we just 'bag' all this nasty proprietary software?  I think
    that would be sad...because I think you still need some software
    'goodies' to help sell your hardware...hell without it your hardware
    competes on a totally level playing field.  Hardware margins being the
    way they are we had better plan on being a 40,000 person company real
    soon!
    
    - we DO need to pare down the software offerings...and what we deem as
    essential we should make available on all DEC platforms as well as some
    key competitors platforms.  But I don't know if our upper management
    has the intestinal fortitude to do that.
    
    - Rdb is getting more portable every day...indeed we do start field
    test this month on OSF...using the same code base as VMS.  So much of
    that wonderful peoformance and availability will go to OSF and NT...and
    could be moved to other platforms in the future.
    
    - small player?  Not yet...last year we were reported by Gartner as the
    3rd largest database vendor worldwide!  IBM and ORACLE were the only
    ones ahead of us.  And believe me our Marketing budget would NOT keep Larry
    Ellison in gas for his car.  We're sliding now with VMS...but if we were
    given the go ahead for other platforms we may be able to pull out of
    the nosedive...alas.
    
    Steve

    -13  Rdb or DBMS for MANMAN
    
    ASK has tried to port MANMAN from DEC DBMS to Rdb and INGRES.
    Application design keeps them with the faster database DEC DBMS, which
    is a CODASYL rather than Relational database.
    
    DBMS keeps getting faster along with Rdb since both Digital databases
    share a large portion of code. DBMS in most applications will be faster
    than even Rdb. It is the pick of many very large corporations for
    stable production applications.
    
    It may be of some interest to some of you to find that Digital's DBMS
    is used to produce chips around the world.  "Intel inside" is with the
    help of DBMS.  FAB lines for Intel, AMD, and DEC are controlled by DBMS
    while producing 386/486/Pentium and AXP products.
    .
    
    

	
>>	OSF and NT are not viable platforms either.  OSF is probably
>>	seen as proprietary to DEC  and NT is not yet a player.


	Hell, let's lock up the store and call it quits then, we've
	obviously got nothing to sell. The VMS Oranges are going bad,
	the NT Apples aren't ripe yet and the OSF Bananas are grown
	from the wrong tree.

	Are you just having a bad day or are you always this depressing?
	VMS is still paying alot of bills around here and there is a good
	market for us (VMS) to capture the ever increasing mainframe 
	downsizing market.

	All is not lost Glum.. We're not doomed yet.


						mike

	I too, wish to echo congratulations to the Rdb team.  

	But, sadly and I hope this does not apply to a majority, 
	there is one segment of this company which needs to be
	sold on Rdb....the Sales Force.   I have met too many sales
	folks who focus sales with an Oracle platform and not Rdb.
	There may be hundreds if not thousands of applications using
	Rdb, but, where are they in the market place ?   Heck, internally
	we're buying products which use other vendors db products.  What
	message does that give ?


	Again, my congratulations stellar product and product team.

	Tom 

    
    
    	Interesting to note that Bob Palmer came to Spit Brook yesterday to
    visit Database Systems...and when someone mentioned that internal
    groups were considering buying other databases he was quite surprised
    and was miffed enough to have his chief of staff writing a lot of
    notes!
    
    

Ref: .1, .2, .4, .9

	Thanks for the good words. We will not let your confidence in
	us be in vain.

Ref: .3, .5

	Add one to the DBMS world, Siemens in Germany produces all their
	chips with DBMS as well. DBMS has over 7000 licenses and still
	makes money for DEC.

	SAP - I have commitment from DBS management and VP level to get
	SAP to port their product to Rdb. Now to convince them that they
	can make a lot of money with us. Platforms are an issue but we are
	working on that.

	BTW, there are over 450 tools and applications from more than 350
	third parties whose products run on Rdb, some of them only on Rdb
	but not many I'll admit. But hell, that is what competition is.

Ref: .6
	Getting depressed Wlodek? :^) Cheer up, we'll work this out or go
	grow carrots and drink bad beer somewhere.

	There is a market for us and we hold a sizable portion of it already
	and will get more if we get our thumb out. As for money, Rdb alone
	made $116 million last fiscal year from license revenues alone. This
	year is expected to be less as VMS is (as we know) getting less as
	well. But even without marketing we are one of the most successful
	software products/projects in Digital and we're going to stay there.

	Confidential? Hell no! This information was part of today's
	announcement at over 100 locations worldwide outside Digital. I also
	have put together a very busy slide (to catch people's eye) and a
	fact sheet containing the technical information from Steve's mail
	which will be given out at German DECUS next week and at the Codd
	and Date DBWORLD conference in London next week. We have a big booth
	at DBWorld this year with lots of hardware and signage which should 
	put our efforts last year to shame. 

	I'll put this small handout in ZUSE::RDB$KIT:[SLIDES]RDB3692.BCK.

Ref: .10
	As mentioned, Rdb will start field test later this month on OSF1
	for AXP. Is this a viable market? Considering the number of requests
	for information and to join the field test I have had I believe we
	have the beginnings of a major breakthrough into this market. 

	BTW, VMS may be slipping in the marketplace but a lot of people have
	and still are betting their business on it. And remember that ORACLE
	gets their best numbers with ACMS and VMS so far.

Ref: .11
	Porting off platform - I agree, we must do that. Period. If we 
	don't we're dead. But we need to make sure people pay attention 
	to us now while we're getting there. I'm betting my future on this
	so you can be sure I'll fight for it. Which ones to port to is the
	question.

	BTW, performance IS everything otherwise people wouldn't be publishing
	these numbers all the time. We can beat every other vendor in almost
	every area but tools and applications in Functionality, SQL compliance,
	Availablity, Very Large Database, Database management, and Security as
	well as Performance.

ref: .13
	Clusters on OSF are coming and we will be the only ones to be able 
	to support them since the technology is DEC's and we've always FULLY 
	supported them. NT is coming and if you don't know that then you are 
	not talking to the right customers. I visit and talk to over 200 
	customers a year and I'm getting a much different story. 

	Also, if I am not mistaken MOST vendor's UNIX offerings are their 
	own not some generic O/S. I also believe we are going to call OSF1
	DEC Unix once the name becomes open to public use. Correct me someone
	if I am wrong on this.

>	The db market is large but Rdb is an insignificant player there.
>	The fact that the benchmark was obtained on OpenVMS is strike
>	three.

	Wrong. We are not insignificiant, sorry. Don't know where you get
	your information from but I think you need to get updated. And as I
	said before, a lot of people put a lot of their strategies into VMS
	and a lot of people get their best numbers and production systems
	from VMS.

>	Either you make software that runs on all the major platforms,
>	or you make platforms that run  all the major software. Or
>	you fill the niches.
>	Trying to be a major player in both only causes trouble now.

	I have to agree on this one. We do have to get onto other platforms
	but we cannot and will not abandon our own.

Larry

.18 - Right on Mike! 

.19 - That is true to a great extent. But we're working on that as well. Talk
to the UK Salesforce, they are pumped up about our DBS products and are even
helping to staff the DBWorld stand next week and do on the floor presentations.
We did Software Sales training here in Europe a while back and some of them are
coming around. Part of the problem is that they pay nothing to get an ORACLE
technical person to help with a sale (but they get less of a profit on the
sale too but that does not count) but have to pay (internal DEC money) to get
an Rdb person. Got to fix that.

.20 - We need to make sure he hears a lot more as well. 

    
    fyi, the market capitalization of Oracle Systems is larger than
    that of Digital.  Amazing, but true.  Just one slice of the 
    database software biz is bigger than DEC, and it's a growing,
    high margin business.
    
    viewing Rdb as a means of improving Digital hardware sales is the 
    conventional (and status quo) view, but it obviously isn't the
    only view.
    
    we're sitting on top of the #1 package in one of the hottest growth 
    industries around, (at least #1 in performance).  now we need to
    decide if Digital wants to be in the business of selling software
    on its own merits.
    
    we aren't well organized to market and sell software, and we are
    very dependent on Oracle, Sybase, ..., to get into customer 
    accounts to sell Digital hardware.  And we need that money.  I
    expect that there will be quite a bit of drama at Palmer's level
    as they chart Digital's direction on this, but from where I sit,
    the choice is pretty clear: expand growth businesses aggressively.
    
    btw, IBM does exactly this with DB2, and it is a very profitable
    business for them.

    Re .several
    
    My conratulations to the RDB folks. Also I think you're dead right to
    sing your praises and be very proud of yourselves. I believe that
    Digital in general won't return to profitable growth until there are
    more examples like .0. Ie people that are passionate in what they do
    and willing to evangelize their success.
    
    May OpenVMS continue to success
    May DEC OSF/1 establish itself as the best UNIX bar none
    May Alpha become the platform of choice for NT and Nt become a
    significant force.
    
    Let's let every "division" in the company do their damndest to be the
    best. But let's stop trying to score points against each other.
    
    I'm beginning to have hope in this company again.
    
    Both of the business that have been spun off as a pseudo division (PCs
    and Storage) are doing well and growing. Pity Storage screwed up their
    manufacturing load plans but at least backlog fulmillment is a "nice"
    problem to have.
    
    Larry Walker now heads up the Network Infrastructure Business segment
    under Lucente. My bet is that if they are successful in building the
    indirect channels Palmer will spin that off from Lucente's salesforce
    and make on a par with PCs and Storage.
    
    Meanwhile we appear to have a Software Business again (segment under
    Lucente). Bud Enright should be given some time (but not too much) to
    prove we can be profitable in software again. Now along with that let
    Lipcon go off and make VMS something to be proud of again in the market.
    Meanwhile Willy Shih can be building a hopefully growing UNIX and NT
    business.
    
    One the plus side the company is now structured in a reasonable way.
    I hope the Board is willing to allow Palmer his one major screwup
    (the CBU structure, now dead). And I hope Palmer is able to recover
    from the general screwup of squandering the year+ lead we had on Alpha
    chips. Things are beginning to be put in place corporate structure wise
    to be successful. We can all help to make it successful.
    

	re: .18

	Yes, I am having a bad day. Has something to do with management
	being unable to read the writing on the wall despite the
	1500W flood lamps and flashing neon signs.

	I've never thought that we didn't have anything to sell.
	I've always been very impressed with our products (granted
	that there are some we'll just forget about).

	I'm afraid that I've just heard and seen too many screw ups
	in getting customers to see these benefits.

	Just one more log on the fire, a friend who is a customer,
	raves about Linkworks. I personally know nothing about it,
	but he seems to think it's hot stuff, nothing quite like
	it in the industry. Unfortunately, he sees no push to market
	it at all and as a result can't get it accepted at his site.

	Just one more turn of the screw on a screwed up day. 

Being a frequest visitor to DB/AD Symposiums, I can't pass this string up...

When it comes to being the multi-platform database of choice, Oracle seems to 
have this pretty well in hand.  Especially when you consider the markets 
current "multi-platform" of choice, Unix (Oracle has 50+% of the Unix market).

When it comes to being the client/server database of choice, Sybase seems to 
have captured the imagination (and dollars) of a good size chunk of that 
market.

When it comes to high-end, production strength database of choice, its probably 
something on a mainframe, maybe DB2 or IMS (non-relational).  This seems to me 
to be a "niche", a HUGE niche, that Rdb running on VMS (Open or otherwise) 
could thrive in!  Most mainframe types are scared to death of moving their 
critical applications to a client/server environment with a Unix server.  Sure 
they can save money, but how safe is it, how solid?

What we have is a production strength environment, with a Unix size price tag 
and the capability to move an application from timesharing to client/server 
gradually as their comfort level rises!

I don't feel we can ever compete with Oracle and Sybase on their turf, that 
window has been closed.  What we can do is recognize a portion of the market 
that plays to our strengths, one where the window is opening wide, and stake 
our claim!  Hell, we might even be able to charge more for it!!


Rick C

P.S.:	Based on the split between Microsoft and Sybase, NT may be a
	new market where Rdb can carve out its niche on even footing.

    
    Great, kudos to everybody, but I have seen this film before.

    Consider :	GIGAswitch - unique product , datacenter "backplane" of the 90'
    		DECNIS     - highest performance IP router
    	        DECbrouter 90 - CISCO software in unique package 
    			        ( hub and standalone)

    		Alpha 		-  ...you fill in ...

    You can add plenty similar things over last few years, and so what ?

    The router market is 1,5b$ , how much of it do we have ?
    We have a patent for branch router but no products.

    RDB + TP blabla - no problem, do you see more then 100 Tbyte databases
    			worldwide looking for replacement ?
    Get all 100 at "Unix price tag" - what have you achieved ?

    We have lots of point successes "and strategic investments" ( or was
    it "tactical losses" )?

    My point is that we should not run internal self congratulations orgy 
    ( as with Alpha), there will be time for it when $$$ start rolling in.

    I thought I made the point (and others did too) that the dollars are
    rolling in - as a result of similar previous announcements similar to
    the one that I put in .0 Unfortunately we have been extremely coy about
    our successes so that seemingly large numbers in our own company don't
    know about them. Ask the account managers of the Milan Stock
    Exchange,LIFFE,PFIZERS,LEGO how valuable Rdb is to the account and how
    much revenue we generate from the database (forget about just the
    license fee). This whole company seems to spend more time knocking
    itself and downplaying it's successes then going out and selling itself
    and it's products. 
    In Rdb we have a good story, as good if not better than all the other
    players in the market. OK we are not everywhere we would like to be.
    
    Yes we need to be on every platform and if you talked to the
    engineering group you would be unlikely to find one person who
    disagreed. We are not being held back by technical difficulty.
    
    I happen to believe in what the Rdb Engineering group is doing and
    think we should make a big noise about it.Now we have the goahead to
    make a noise about this. So go do it.
    
    /phil 

    Re .5
    
    Have you heard of ACMSxp (eXtended Platform) and MIA.
    
    Yes, we need to get our TP and DB software on multiple platforms, but
    it is happening - OSF first, of course. When it gets there it will,
    have a fight, but the base article shows that Rdb will have a chance to
    compete. 
    
    Re .21
    
    Two more for DBMS - Plessey and National Semiconductor. 
    
    Re .27
    
    Oracle used to be the main relational database on VMS. Rdb beat it
    there so it must have a chance of winning on OSF, and then, hopefully,
    going on to further success.
    
    Peter

re: .19, .20

When a group in Digital is looking at software, should the selection criteria
be the product that helps the group perform its job the best it can and be
more competitive in the marketplace, or should the selection be limited to only
those products that use RDB?

Bob

>When a group in Digital is looking at software, should the selection criteria
>be the product that helps the group perform its job the best it can and be
>more competitive in the marketplace, or should the selection be limited to only
>those products that use RDB?


Good question. But did you ever think that those groups within Digital might
act like a real customer and ask the vendor to get his product ported to Rdb if
Rdb will do the job better (not to mention being cheaper for us and support if
much more widely available to you).

What gets me mad is when Digital groups won't even consider Rdb over another
database vendor when we do have everything they need to be successful and
competitive.

Larry

re: .32

>Good question. But did you ever think that those groups within Digital might
>act like a real customer and ask the vendor to get his product ported to Rdb if
>Rdb will do the job better (not to mention being cheaper for us and support if
>much more widely available to you).

I would hope that each group would ask, however, at the same time, if I was a
vendor, I'd want more than just Digital asking before I considered a port to
RDB or any other RDBMS.

>What gets me mad is when Digital groups won't even consider Rdb over another
>database vendor when we do have everything they need to be successful and
>competitive.

I'm not sure I understand here.  Are you saying that there are software products
from 3rd parties that run on multiple RDBMS and when given the choice of which
version to use, groups within Digital specify the non-RDB version?

Bob

As of today (April 13, 1994) the 3692 TPS announcement is PUBLIC knowledge
and is probably being flooded around the various internal AND external networks
& mail servers as I'm typing this...

>When a group in Digital is looking at software, should the selection criteria
>be the product that helps the group perform its job the best it can and be
>more competitive in the marketplace, or should the selection be limited to only
>those products that use RDB?

Good question.  Obviously, the product that solves the problem is the product of
choice.  If the need is IMMEDIATE, then there's no question.

However, the need is seldom immediate, so several options are available.
If the product runs on Rdb, then it's a simple matter.  If the product runs on
VMS, then it can be easily ported to run on Rdb (at OUR request!!!).  If the
product only runs on non-VMS platforms, then its (probably) running on a 3rd
party database product and can probably be easily ported to any VMS platform;
again, the solution depends on how much faith we have in our own
platforms/products.

Rick

    I would like to see groups in Digital use Digital products when ever
    possible. If for no other reason then that they could provide immediate
    feedback and enhancement requests to development groups. It's helpful
    and desirable to have feedback from people who use products for real
    work. Besides, if we can't use it internally why should a customer?
    Better we should use it internally and make sure that it performs well
    and meets real world needs before we foist it off on customers.

    I remember talking to an internal MIS group many years ago about
    using VAXes. They said that the VAX wasn't ready yet (it had been out
    for 3-4 years) for commercial use. They weren't telling anyone what 
    they needed the VAX or VMS to do differently though. Is it any wonder
    it took so long to get it accepted externally?

    			Alfred

	As one of those MIS types that has been involved in many of the
	decisions being discussed here I can honestly say that in the
	past much of MIS has bent over backward trying to use DEC products.
	It has only been in the last year or so that a major change of
	attitude has come about.

	There is one problem that has rarely been mentioned when discussing
	using DEC software in internal apps. That is:
	MIS is constantly being pushed to incorporate the latest technology
	and the latest technology usually can't cut it.  How many of
	you would like to bet your business on V1 of any product?

	MIS has been burned repeatedly  by failures of over-optimism
	with respect to new technology. The doom and gloomers were
	proved right with Rdb, ACMS, TDMS, DECforms, Datatrieve,
	Rally, etc. By the time these became "industrial strength" a
	large amount of resistence had been developed and people became
	unwilling to go out on the same limb again.

	There is problem with feeding back to engineering our problems
	because the cycle time to get appropriate fixes is much longer
	than the cycle time needed to get the application working.
	This is no one person or groups fault. It is just a fact.

	re: .34

	Contrary to your opinion the need is most always immediate!
	For a variety of reasons, the pressure is constantly on to
	upgrade to the latest techno whizzy that some manager saw in 
	some trade rag. Our list of failures at these attempts come
	from MIS failings, business failings and the failings of
	our software base. And as a result we wind up with creaky 
	systems that every user hates. With the pressure on the company
	now to downsize and be more efficient it is seen as much cheaper
	and faster to purchase the solution than to write it ourselves.
	We have to go buy what's on the market now, we can't wait until
	SAP decides to port to Rdb.

	With the ongoing rumors as to what software will or won't survive
	the shakeout it is easy to understand the reticense of IS to jump
	on any particular product.

	Just to give you an example, when I mentioned this great
	TP announcement to someone senior to myself, I was told that
	Bob Palmer had forbidden us from releasing the information
	because Rdb was on shaky ground because of our burgeoning relation
	with Oracle.  Now, how should I react to that?

>	Just to give you an example, when I mentioned this great
>	TP announcement to someone senior to myself, I was told that
>	Bob Palmer had forbidden us from releasing the information
>	because Rdb was on shaky ground because of our burgeoning relation
>	with Oracle.  Now, how should I react to that?

I heard something similar.  Could somebody verify or debunk this?

    In the field, NY in this case, we have been told to work closely
    with Oracle and Sybase for new database sales.  The rationale is that
    we have been losing system business by not working closer with these
    DB companies.  A striking statistic showed that Digital had lost a
    significant % of system sales for these products.  I don't remember
    the exact numbers, but it was something like a few years ago about
    20% of these DBs were on Digital systems.  Now it is more like 5%.

    
    
    RE: Palmer forbiding discussion of the benchmark?  Horse manure.  If
    your friend would like to he can give me a call.
    
    RE: Rdb and the Database Market:
    
    	- When I started with Rdb in 1987 everybody said 'Ingres has won,
          you are wasting your time on VMS'
    
    	- In 1988, that changed to 'Oracle has won on VMS'
    
    	- In 1990 Ingres and Informix had won on Unix...
    
    	- In 1991 Rdb was the only real choice for VMS according to the 
          analysts...
    
    	- In 1993, what happened to Ingres?  Gee, Informix looks like they
          are coming back from the brink...
    
    	- So now in 1994, Oracle has won Unix, Sybase client/server etc etc
    
    
    See why I don't think it's all over?
    
    

    Re: Palmer forbidding the release of the performance numbers  It's bunk
    
    	Unless he is talking out of more than one side of his face, Palmer
    told the assembled Rdb engineering group that "he supported releasing
    the numbers" but did note that some of his direct reports didn't.  
    
    	You must be hearing from his VPs who apparantly don't speak as
    Palmer does.

re: a few back.

Yes, groups in DEC have chosen non-DEC software when we had what it took to get
the job done, were cheaper, faster, more functional and the tools were
available from inside and outside of DEC.

Re: MIS managers woes.

I don't understand why you can't make it work. I've got hundreds of customers
around the world who are making it work and work well with Rdb, ACMS etc.

LIFFE in London trades billions of Pounds Sterling a day and makes tons of
money and they are a VAX shop running Rdb. On black friday or whatever the day
was when the Pound nosedived they did over 800,000 (no typo) contracts in one
working day with no hicups.

The Italian Stock Exchange is another one, they have advanced their electronic
stock exchange from 10 stocks to 200 in 12 months and are going for the entire
shooting match in the next 12 months. 

These people and others have been making it work for several years. So at what
point do you think Rdb became 'industrial strength'? In those days no one was
and we have beaten all the others in several cases and for more than just the
last 12 months.

So what do you mean by we failed with Rdb, ACMS, etc in the past? Who failed?

Larry

    re: 36
    If you have been burned repeatedly in the past by using new technology,
    OK, I can understand that.  Admittedly, Rdb had serious quality
    problems around 3.0, years ago. But there are 5 or 6 versions of Rdb which
    are out there and supported currently.  All the necessary bug fixes
    are being built into new releases of those 5 or 6 versions.  Surely
    one of them must be stable enough for your group.  They're stable 
    enough for Ford, the London stock exchange, the largest Swiss banks,
    etc...  The service has been voted best in the industry by large
    database customers.
    		Tom

re: 41

	The failures go way back to the early '80s. When Rdb V1 couldn't 
	handle the load
	necessary to run the Price/Product System, when ACMS V1 couldn't
	be made to do what was needed to re-write the order entry system
	etc. etc. the "new-technology-phobes" were "proven" right.

	The problem is that we in MIS find out about all this stuff
	TOO SOON! If you've been hearing about this great new product
	for a year or so, it becomes hard not to use it. But, what
	you were hearing were from the X and T versions. So when 
	V1 finally comes out (we could usually hold off that long)
	people expected a product up to V3 or so level of maturity. 

	Digital's internal applications are larger and more complex
	than most of our customers. We COULDN'T run with the level
	of quality and "doneness" of the V1 products. That is not
	necessarily anyone's fault. BUT, by pretending we not only
	could, but, had to; we tried and failed and made the skeptics
	right. So, when time had passed and the products improved
	they couldn't be sold internally with out a struggle. Here
	all of the MIS internal politics comes into play.

	Now, we are faced with a totally new issue which is, we can't
	write it as fast as we can buy it. So, what's on the market
	and we don't care if it uses DEC software as long as it runs
	on DEC hardware (though even that is no longer a given).


	

    
    .37, .39, and .40 are the most important discussion points in this
    entire string -- with the exception of the base note (IMO, of course).
    
    If there is a debate at those levels, it needs to be put to
    rest quickly. The strategy needs to be communicated to 
    sales/consulting professionals. Now.        
    
    Glenn

    re: .43
    
    I worked in MIS in the early to mid 80's and my observations were that
    we used the best product for the job. At one point, we were evaluating
    1032, Rdb and Powerhouse for the services data warehouse and we went
    with 1032 as Rdb couldn't do the job at the time (determined by our
    benchmarks). We did wind up shelling out big bucks for the external
    licenses. This was in the V2.? timeframe.
    
    I moved over to database around the time that 3.0 shipped and sometime
    after that, we had many internal projects interested in Rdb. We had
    many internal products that did go with Rdb. It's just lately that
    we've had the issue with buying an external product due to layered
    software issues.
    
    With database software, customers make an investment and then usually
    stick with it for a while as it's hard to justify throwing away the
    large initial investment. Decisions can also be emotional, such as:
    I know Oracle/Sybase/Informix/Ingres/Rdb and I'll recommend it or favor
    it in evaluations.
    
    I'd say to Digital's MIS that they give us a chance to make our
    case. Let us know if they have complaints as our external customers do
    so that we can fix the problems.
    
>	right. So, when time had passed and the products improved
>	they couldn't be sold internally with out a struggle. Here
>	all of the MIS internal politics comes into play.
    
    Same thing happens at customer sites. Wastes a lot of time and money
    for our customers and probably for us too.
    
    michael

    Re .43
    
>	Digital's internal applications are larger and more complex
>	than most of our customers. 
    
    They are also of much poorer quality than most of our customers, even
    the ones using even larger and more complex applications.
    
    When I was a customer, we had a policy of not using .0 versions of
    anything for production. We would always wait at least six months to
    see if a .1 version was coming.
    
    Peter

    
    >	you would like to bet your business on V1 of any product?
    
    Are you saying that our customers shouldn't buy V1 of any product
    we sell?
    
    			Alfred

re .43:

>Digital's internal applications are larger and more complex
>than most of our customers.

I seriously doubt the validity of this statement.  

When Intel and Natl Semiconductor use Rdb (or DBMS) to build chips and DIGITAL
doesn't, what does this say about our internal processes???

Rdb currently has customers whose complex applications are the life-blood of the
business - they cannot afford to go down - and they choose Rdb (or DBMS).

Let's face it - Rdb succeeding is good for business!

Rick

Steve Hagan has asked to have this posted in this note.

From:	NOVA::HAGAN "Steve 381-2425 Engr Mgr INDUSTRY LEADING Rdb + DBMS  12-Apr-1994 1229" 12-APR-1994 12:33:21.00
To:	@GROUP.DIS
CC:	HAGAN
Subj:	Palmer Congratulatory Visit to Rdb Engineering


+-+-+-+-+-+-+-+
|d i g i t a l|         I N T E R O F F I C E   M E M O R A N D U M
+-+-+-+-+-+-+-+

TO: Rdb/DBMS Engr Group
	     				DATE:  April 12, 1994
                                        FROM:  Steve Hagan
                                        DEPT:  Rdb/DBMS Engineering Mgr
					EXT:   381 - 2425
        				LOC:   ZK2-1/O.C.1 
        				ENET:  NOVA::HAGAN
       
cc: Rozwat

SUBJ: Minutes from Bob Palmer's Congratulatory Visit to Rdb Engineering

Bob Palmer visited the Rdb engineering group on April 11, 1994,
to congratulate the group on TRIPLING the WORLD RECORD for tpsA,
and update the group on the planning process and the current status.
He spoke for about an hour, with extensive Q/A.

This was a phenomenal compliment to the group! 
Some examples are (not well known stories) (my paraphrasing)

	Palmer had not even been the the Spitbrook facility in 4 YEARS!.

	This was his first time speaking to a software group.

	Palmer complimented the group as having built the best software
	product in the company.

	Palmer spent about 2 hours on this meeting, including driving time.

	Palmer appointed Rose Ann Giordano to develop a software strategy, 
		working with Mark Chardon, while at the meeting.
		This gives us the opportunity to work closely with Rose Ann
		to determine our future. The Rdb strategy is a good one,
		we needed the forum to present it to Palmer, now we have it.

	Palmer is well versed in all the debates + options on company 
		strategy - the "only box" shop, the "software as a business"
		tradeoffs, the "silicon, service, + software" combination -
		which has many suboptions, the pros + cons of both 
		cooperating with and competing with 3rd party software 
		houses.He understands the "installed base" issue + 
		the effects on revenue of the possible options.
		He is well "in touch" with the issues; he is NOT remote or 
		removed.

	Palmer believes that software houses work with us because they
		make money doing so, and are minimally affected by whether 
		Digital has its own competing software package. Their 
		interest is financial: if Alpha is fast + popular enough, 
		3rd parties will make a profit running on Alpha and will
		do so. Otherwise they will not. Straightforward.
		
	Palmer seemed very honest and open in his answers to all questions.
		No sugar coating, no beating around the bush.
		One clear message was that people + products are expected
		to generate revenue and profit for the corporation. 
		Otherwise both will disappear. 

		This is normal business, as practiced in most companies. 
		Too many people at Digital seemed to have adopted the 
		"You owe me lifetime employment" model of day to day 
		activities. These are the ones who are costing the rest 
		of us the problems we face.
		
		A clear message is that Digital does not as yet have a
		software strategy that is agreed to by all his reports. 
		Even products as successful as Rdb need to articulate 
		their plans, position themselves with the 3rd parties, and 
		show why the corporation will receive a good return on its
		investment.

Steve Hagan
Rdb Engr Manager

	Bob may not be remote or clueless but I'd bet it would be 
	difficult to say the same about some of his direct reports based
	on some of the stuff in this notesfile and "thru the grapevine".

							mike

    First off - congrats to the team!
    
    I would like to ask a question regarding competition, market share, and
    specifically, Oracle.
    
    I work in the US Third Party Acquisition organization and for a long
    time, the Third Party Field teams have incorporated Oracle products
    into solutions we've sold when for whatever reason, Rdb wasn't the
    Customer's choice.  About a year and a half ago, I was asked to support
    what was then the US SI organization in negotiating a master distribution
    agreement (MIDAS) for Oracle's products.  We spent 6 months in 
    neogitations and came away with nothing.  My impression as to why that 
    happened was because neither company was viewing the relationship from 
    the same perspective.  Oracle, it seemed, wanted Digital to be its VAR 
    (complete with a dedicated organization to handle pre-sales configuration 
    and post-sales technical support as well as other VAR responsibilities).  
    Digital just wanted the ability to distribute Oracle's products as 
    required in the simplest manner possible (I believe at that time we
    were distributing some $2M+ in Oracle products annually through the US
    alone - don't quote me on that figure though...).
    
    To this day, I'm unclear on exactly what we as a company want to or
    don't want to do with (or against) Oracle.  On any given opportunity,
    we may be eachother's customers, competitors, re-sellers, etc.  How
    friendly do we want to be with them and under what circumstances?  Do
    we view ourselves as competitors, partners, both, neither?
    
    To try to make this a short story, let me just say that I've been asked
    to conduct a review of our relationship with Oracle as a Supplier of
    S/W products to Digital; and of course, from my perspective, part of
    that review must focus on what they supply to us for resale to our
    Customers - but that is only one facet of the relationship and I would
    appreciate any help or pointers you can provide.
    
    If anyone here has any relevant information on this, or any other
    aspect of our multi-faceted relationship with Oracle, I'd really
    appreciate hearing from you (preferably off-line, but I'll continue to
    look here).
    
    Second, if anyone can point me to market share information on DB
    vendors, that would be helpful.  The SI organization I was supporting
    at the time had figures that showed Oracle to own more than 50% of the
    total DB market (can that be right?) and Digital showed up 5th in line. 
    What is our goal for DB market share and what is our strategy for
    getting there?
    
    Lastly, if anyone has any light to shed on how Digital as a whole or by
    segments views our relationship with Oracle, that would be helpful too.
    I guess personal opinions are welcome too; especially as to why our
    Customers may be asking for Oracle first and what we can do to change
    that...
    
    Thanks for whatever information you can provide, and again,
    congratulations to the Rdb team on a fantastic technical achievement!
    
    Steve McConnell
    POWDML::SMCCONNELL
    DTN 223-7709

    
    
    RE: -1
    
    I'll forward some postscript slides on DB Marketshare...
    
    -Steve

re: .47

	No, I'm not saying that no customer should buy our V1 products.
	I'm asking if you would be willing to bet the integrity of
	your revenue stream on a V1 product.

	As I've said before, I like and respect our software products.
	I have used many of them and have seen some of the teething
	problems that the early versions have had.

re: .48
	Notice I said most not all. You would be hard pressed to find
	too many 10+billion dollar companies running all their admin.
	systems on VAX clusters with networks. Our admin. systems are
	large enough to tax most of our software.

	I'm glad that Rdb is succeeding. But, could Intel have run their
	manufacturing on Rdb V1 circa '83?

re: .51
	Interesting timing. I just was in a meeting with 3 folks from 
	Informix trying to convince us to use that instead of Oracle
	underneath our R3 system.  They claim to have the largest share
	of the Unix Rdbms market by number of licenses while Oracle
	has the dollar lead.  Rdb isn't a player in this space yet, which,
	in my humble opinion, is too bad.

>	No, I'm not saying that no customer should buy our V1 products.
>	I'm asking if you would be willing to bet the integrity of
>	your revenue stream on a V1 product.
    
    I would not ask a customer bet their business on a product that I 
    wasn't willing to bet my business on.
    
    				Alfred

    

    Point 1 .	This great achievement just opens a possibility, it is a 
    		condition for success. There is nothing more to it.

    		How much money did the test cost ? How much money does RDB
    		development cost ? Is marketing money spent on marketing 
    		RDB .5%, 1% or 50% of the development budget ?
    		How does that match what Oracle and Sybase spent on
    		marketing ?

    Point 2 .	How to lose RDB sale.

    		1-2 years ago my customer was looking for a DB system It is
    		VMS only shop with OS/2 front end workstations. Since
    		lights out computing ( remote management ) and very high
    		availability was required, we were pushing VMS/RDB.
    		Local sales and account team were rather reluctant to do 
    		presales song and dance, they have lost too many sales like
    		that on the Wall Street. Anyway, I've achieved to change
    		their minds and give it a try. We had also some allies
    		within the company that wanted VMS/RDB.

    But they have picked SYBASE and OS/2 servers. And problems started.
    IBM M95 was too slow and small , application was split to second
    system. You can't remote install the stuff, 3 people had to go for
    weeks to Tokyo to install it. Forget any reasonable backups ( 200+
    diskettes) .In the case of a crash one needs to scratch install OS/2,
    Pathworks, sytor software and only then rest can be installed. THE BIG
    MESS . Obviously wrong decision. But, if they had to take some
    decision today, they might have picked same solution again.

    Why ?

    In this company, as in many other companies, lots of work and
    decisions are made by external consultants. What is consultants
    objective ? To make buck on current project and learn skills usable for
    the next assignment. Just ask yourself how many RDB/VMS projects are
    there at the Wall Street as compared to Sybase/OS/2 ?
    ( BTW. NASD news came to late for us)

    So, unless there is a massive marketing effort showing that RDB is used
    and RDB consultants have jobs , we have lost in advance. It does not
    matter how many times we beat world records.

    BTW, an interesting observation : the massive use of consultants creates
    different then before market dynamics. There is less room for taking
    technology risks and not-reusable solutions. The main stream gets
    stronger by avalanche effects.

    DEC has to decide, play main stream ( invest $$$) or play niche .
    

>    DEC has to decide, play main stream ( invest $$$) or play niche .

	That just about sums it up. Right now we invest just about ZERO in
	marketing our software, and ZERO in advertising it. Heck, take a
	look at our LIVEWIRE and Internet (Biz.Dec.announce) articles about
	the April 12th announcement. Do you see the world record benchmark
	mentioned *ANYWHERE*? Not -- but you will see mention of our
	marketing agreements with Oracle, Ingres, and Sybase.

	This situation is too common in this company. Take an outstanding
	product that is a technical leader by any objective measure, and
	emasculate it by

	a) shackling it to DEC hardware platforms, so it's always perceived
	as proprietary (not "open") by customers and analysts

	b) not marketing it, not pushing it in the market, and not getting
	our sales people even aware of (much less trained in) the product

	The question is, why do we spend so much time enticing third-party
	software vendors to port to our platforms, but actively PREVENT
	our own software products from porting to other platforms? The
	answer is that this is still a hardware-centered company, in which
	the strategy is to ship boxes in volume. And anything that *might*
	get in the way of that (like, "Oh no, Oracle might get mad at us if
	we announce this benchmark") is quashed by some of the most powerful
	people on the SLT. For some reason this strategy is seen by some as
	the route to success.

	What we need to decide is, "Do we want to sell software for profit,
	and if so, how?" The good news is that, according to Bob Palmer as
	of his Monday visit to talk to us, the issues are known, the
	alternatives are being analyzed, and decisions in this space will be
	forthcoming. We can only hope that the decisions will be successful
	for the company as a whole.

	Roy

    	With the rumored big cuts coming to software engineering in Q4, I
    think we're getting a peek at what the decision might be. I don't think
    Palmer or the SLT think we can be successful in the software business.
    I think they'll use the mis-managed past as evidence. I think that the
    only software the company will invest in in the near future is that
    which is necessary to sell hardware (i.e. Alpha chips in particular).
    The writing seems to be on the wall for this although I think it is a
    losing strategy. Personally, I like the *new* system software group
    organization under Don Harbert (all OS's, compilers, tools, etc.).
    However, I think we ALSO need to split off a software arm of the
    company as was done for storage and PCs to give them a chance to be
    successful and let them focus on a few, profitable areas to compete in
    the open market. We've never really given ourselves a chance to do
    this, and I not only think it's worth the risk, but I think it's
    necessary to keep a reasonable revenue and profit stream going.

re .49:

>This was a phenomenal compliment to the group! 

>	Palmer had not even been the the Spitbrook facility in 4 YEARS!.
>
>	This was his first time speaking to a software group.

Indeed it's a compliment to the Rdb group, but it also says a lot about
Palmer's opinion of software.  It's kind of unsettling for us software types.

>	Palmer spent about 2 hours on this meeting, including driving time.

When you consider the driving time, it's not all that much meeting time --
unless he drove the Porsche.

    re: using what you sell
    
    Scott McNeally (sp?) the CEO from SUN has been quoted as saying that
    they'll be turning off their last mainframe (IBM clone Amdahl?)
    sometime this year.  At this point they'll be running a peer to peer
    (n.b. Client/Server *NOT*) network for their entire business.  Now that's
    confidence in what you sell.  He claims many of his customers want to
    do the same, but he is *LEADING* by example.  Their last financial
    stumble was when their mainframe MIS systems were upgraded.
                            
    [Rathole alert:  at last count, SUN has been profitable for more quarters 
    than I can remember and are heading to becoming a 4.5B company this FY 
    up 200M while we've shrunk 734M in the 2 quarters to date]

    Re:
    
    Handwriting on the wall, going out of the software business?
    
    If this is how you see it, how do you explain the US Field
    INSISTENCE on filling software sales specialist jobs all
    over the US?
    
    Sounds like we'll have people desperate to sell our software just
    as the group that builds/supports it goes away!
    
    Classic Digital (DEC)
    
    
    Dan

    re: the importance of software for revenue
    
    If you do the numbers and add up the combined revenues of the software
    top-10 or so...
    
    Microsoft, Novell/WordPerfect, Lotus, Borland, Oracle, Sybase,
    Informix, Ingres, ComputerAssociates, ???
    
    You'll probably get less than $10B in revenue.  We are probably better
    off concentrating on PCs (we have ~2% of $66B), Disks, Workstations (we
    have ~10% of $10B) and Midrange (we have ~14% of $21B)
    
    Hardware seems to be where the revenue is, although margins are tight
    compared to software.  [having said that, the rash of mergers in the
    software area and the price wars, we may be seeing the shrinking of the
    software market, at least in PCs]

re .61

Statistics of packaged software sales:

Rank   Vendor		1988	1989	1990	1991	1992	1993

1	IBM  		7,927	8,424	9,952	10,524	11,103	9,963
2	Microsoft 	  635	  829	1,252	2,019	2,830	3,575
3	Computer Ass. 	1,305	1,484	1,526	1,589	1,590	1,845
4	Oracle    	  350	  668	  778	  800	1,058	1,392
5	Novell    	  -	  -	  456	  640	  889	1,064
6	HP		  500	  550	  600	  702	  794	1,014
7	Digital		  750	  850	  940	  967	  988	  952
8	Lotus     	  437	  534	  547	  716	  796	  846
9	Sun  		  110	  150	  425	  531	  590	  692
10	AT&T C/S	  349	  408	  575	  623	  676	  684
11	Unisys    	  650	  700	  800	  740	  712	  655
12	WordPerfect   	  179	  281	  452	  532	  532	  577
13	Bull 		  -	  -	  684	  662	  598	  544
14	Borland   	  398	  369	  417	  497	  461	  457
15	Legent    	  149	  202	  283	  319	  434	  437
16	Apple     	   80	  90	  290	  329	  370	  433
17	Autodesk    	  114	  173	  238	  282	  368	  418
18	Software AG    	  154	  188	  236	  305	  335	  406
19	SAP  		  -	  -	  192	  271	  289	  400
20	SAS Instit.	  155	  188	  220	  266	  329	  397
21	Sybase    	   24	   57	   89	  133	  220	  355
22	Informix	  103	  145	  133	  149	  238	  328
23	Ask Group 	  190	  238	  264	  305	  352	  322
24	Cadence		  -	  -	  282	  311	  334	  280
25	Adobe     	92	121	  163	  223	  233	  279

Source: Readers Choice, IDC Gray Sheet - February 11, 1994 issue  	
						

>    	With the rumored big cuts coming to software engineering in Q4, I
>    think we're getting a peek at what the decision might be. I don't think
>    Palmer or the SLT think we can be successful in the software business.

	Based on what Palmer told my group, I think this is too simplistic.
	First, let me digress and say that I found Bob Palmer to be
	- eloquent & articulate
	- knowledgeable about all of the issues that were raised
	- very forthright and direct

	He didn't sidestep any questions, and gave honest answers. I can't
	emphasize enough how much better this Q&A was than any I've seen
	with DEC senior management in the past.

	Anyway, it's clear that the current business model isn't working,
	and that choices have to be made. These choices range from

	- disinvesting in softare, and essentially putting our products into
	  maintenance mode

	- spinning off some or all of software into a separate company

	- selling some or all of our software to a third party

	- deciding to copmete as a software vendor, and developign a
	  focused, viable strategy to do so

	Let's face it -- all of these options have problems. The fact is
	that we've done a lousy job of managing and selling software in the
	past. We've also done a lousy job of cutting deals with other
	companies, and we have no experience in spinning off something so
	big as our software products. And those are just top-level issues,
	believe me.

	Will there be major cuts? I think so. There are some products we
	make that clearly can't compete in ANY business scenario. But the
	range of possibilities might include my personal favorite, a focus
	on high-performance, high-quality productions system software. Take
	the remaining image we have in the market (high quality and high
	performance, in PCs, Disks, Alpha, and networks) and develop that
	side of the software business. Do what it takes to compete (which
	means at least a credible presence on third-party hardware
	platforms, if not a full-blown software marketing effort into those
	market spaces), ADVERTISE, MARKET, and SELL. We have the products --
	this benchmark proves it, and it's not just limited to Rdb & ACMS.
	But do we have the will and desire to compete?

	Roy

|    Are you saying that our customers shouldn't buy V1 of any product
|    we sell?


     Because of data integrity issues, no DBA in their right mind would
     ever migrate a production database system to a V1.0 product. 

|    re: -1

     One aspect that has been lost in this discussion is the future
     of database systems.  Ten years ago the wisdom was that very
     large database systems could exist exclusively as a software
     solution (i.e. the notion of database machines was dead.)
     Looking at current research and the drive for massive parallelism
     it seems that there will be a much closer relationship to hardware
     and software.  The bottom line is if you intend to continue selling
     big machines, you better not keep an arms length away from database
     technology.     
  

.55>    How much money did the test cost ? 
    
    Remarkably less than the profit which Digital will receive if we sell
    even one (1) system comparable in size to the tested configuration.
    
    (Posted with the permission of the manager of the group that did the
    benchmark.  She preferred not to release the actual $ figure, but it
    was _remarkably_ less than Digital's margin on a system such as the one
    tested.)
    
    	/john

    
    re: sales figures
    
    	Looking only at sales is misleading.  Looking only at past sales
    	is even more misleading.  I think it is fairly plain that software
    	costs a lot less to manufacture than, say, a CPU board.
    
    	It isn't necessary to value a company by looking in the rear
    	view mirror in this way, and fortunately valuation is what analysts
    	are paid to do; all you need to do is look at the market value of 
    	the company, i.e. the price of its stock times the number of 
    	shares of its common + preferred stock.  This is roughly what 
    	it would cost George Soros if he decided to buy Digital or 
    	Oracle or Microsoft.  That's the best, unbiased estimate of what 
    	a company is really worth.
    
    	By this (much more relevant) metric, software is indeed bigger than
    	hardware.  Anybody who knew differently could make a killing in 
    	the stock market.
    
    /jim   

                      TPC Benchmark (tm) A Test Results as of April 12, 1994 - Top 20 Performance (overall)
                      -------------------------------------------------------------------------------------
|=======|=========|=======================|========|================================|==============|=======|=======================|
|Ranking|Company  |System                 | Date   |Software                        |OS            | Spec. | TPC Benchmark  Results|
|       |         |                       |Pricing |                                |              | Rev.  |    tpsA  |   $/tpsA   |
|=======|=========|=======================|========|================================|==============|=======|==========|============|
|   1   |Digital  |4-Node DEC 7000-650 AXP|4/12/94 |ACMS V3.3, DEC Rdb V6.1         |OpenVMS AXP   |1.2    | 3,692.02 |    $4,873  |
|       |         |       VMScluster C/S  |        |                                |        V6.1  |       |          |            |
|-------|---------|-----------------------|--------|--------------------------------|--------------|-------|----------|------------|
|   2   |IBM      |ES/9000 Model 511 TPF  |12/15/93|TPF 3.1(Transaction Processing  |TPF 3.1       |1.2    | 3,504.93 |    $7,964  |
|       |         |  Uni-processor        |        |        Facility)               |              |       |          |            |
|-------|---------|-----------------------|--------|--------------------------------|--------------|-------|----------|------------|

For those of us with 80-width screens:

              <<< Note 2995.67 by DIODE::CROWELL "Jon Crowell" >>>
              -< IBM ES/9000 = 3,504 TPS -- 3% more, not 3X???? >-

4-Node DEC 7000-650 AXP|ACMS V3.3, DEC Rdb V6.1        | 3,692.02 |    $4,873  |
ES/9000 Model 511 TPF  |TPF3.1                         | 3,504.93 |    $7,964  |
                             

Now the explanation:

TPF isn't a relational database.  It's a system that was written for American
Airlines.It's essentially an operating system that deals with messages/txns.

It's quite primitive.  For example, application code had to be written to
handle the kind of journaling required to make it pass the Durability tests.
It takes many less instructions to do it's txn than a real DBMS.

Ruth
CSG Performance Group

re: .67

I believe that the official memos state 3X more than any 
other *relational* database system. Specifically we are referring
to the 1079 tpsA that  Oracle did on a DEC 7660 and the 1072 tpsA
that Oracle did on a pair of Sequent servers.

The IBM number was done using TPF or Transaction Processing
Facility on a uni-processor mainframe. TPF is a highly specialized
beast. It's the OS, the DBMS and the TP monitor all in one.
It is nothing like a general purpose relational database
system.

BTW: TPF is the software that runs most of the airline reservation
systems. Indeed it was created jointly between IBM and
American Airlines (?).  I doubt that anyone but the airlines
runs TPF. I've even been told that IBM loses money maintaining
TPF. But of course, their airline accounts are profitable as
a whole.


Robert Abbott
Rdb/OSF Performance Engineer

    re: stock market valuation
    
    I think that this is only the *perceived* value of the company and in
    many ways doesn't reflect reality.  
    
    Microsoft has (or had) a greater stock market valuation than IBM.  Do you 
    really think this reflects reality in that IBM has much larger assets 
    (people and otherwise) and a much larger cash flow and potential for 
    profits.  Microsoft has a huge potential to fall in value as its growth
    slows.  So much for the brilliant analysts who buy now...
    
    I think investors try to buy stock where their ROI (return on investment) 
    will be highest through share price increase and/or dividends.  They
    seem to be remarkably short sighted and company valuations go up and
    down all the time on the slightest hint of news, real or otherwise.
    
    In summary, revenue is important (#2) alongside profits (#1).  We need
    to increase our hardware revenues to make profit due to the low margins. 
    If we have some profitable software, that's icing on the cake.  We
    can't support the many hundreds of current layered products we now have
    that *aren't* profitable.  I suspect rdb is though...

    re: .69
    
    Having managed TPF Application Development for a major bank, I want to
    correct some information.  At the time I was with the bank, I went to
    the IBM Management Institute and they said there were 17 customers for
    TPF.  Seven of those were represented in my class and only two were
    airlines.  You are correct that ACP (Airline Control Program) was
    developed for the airlines and IBM changed the name to TPF because they
    thought that some of us in banking would not accept the original name.
    
    We always thought of it as an OS that had been lobotomized!  It really
    didn't do anything, but you could certainly  push tps!

Re: .60
    
>>>    If this is how you see it, how do you explain the US Field
>>>    INSISTENCE on filling software sales specialist jobs all
>>>    over the US?
>>>    
>>>    Sounds like we'll have people desperate to sell our software just
>>>    as the group that builds/supports it goes away!

When I was talked to about my interest in one of those jobs, I was told we 
would probably be getting credit for selling Oracle, Sybase, etc.

Of course I immediately wondered about SQL Server on NT...probably the 
easiest sell of all in an SME type market.

Rick C

P.S.:	By the way, aren't there some software packages we sell that run on
	hardware other than our own?  Do customers consider these packages
	"open"?  Also, I noticed that HP has passed us in total software 
	revenues.  I wonder how much of their $1B in sw revenues came from
	layered products running on non-HP systems.  And why is it they seem
	so comfortable/content being labeled a hardware company who just
	happens to sell a lot of software, while we want to be Microsoft?
	When it comes to selling hardware, I guess "been there, done that"
	must apply.

> Of course I immediately wondered about SQL Server on NT...probably the
> easiest sell of all in an SME type market.

Probably true with Intel and MIPs. I've heard that there's been a lot of tuning
for those architectures.

> By the way, aren't there some software packages we sell that run on
> hardware other than our own? 

Yes. We use them as an example of why we should be on other platforms.

> And why is it they seem so comfortable/content being labeled a hardware
> company who just happens to sell a lot of software, while we want to be
> Microsoft?

I don't think the company believes that it can become a Microsoft. I think we're
more a hardware company that wants to sell software to generate hardware sales.

|   1   |Digital  |4-Node DEC 7000-650 AXP|4/12/94 |ACMS V3.3, DEC Rdb V6.1
	|OpenVMS AXP   |1.2    | 3,692.02 |    $4,873  |

	3,692.02 x $4,873 =  $18M

|   2   |IBM      |ES/9000 Model 511 TPF  |12/15/93|TPF 3.1
	|TPF 3.1       |1.2    | 3,504.93 |    $7,964  |

	3,504.93 x $7,964 = $27.9M


The DEC, ahem Digital, solution is (1) a real database on a real O/S and
(2) 1/3 cheaper.

    ed

    
    re: .70
    
    	then, by all means, leverage yourself to the hilt and buy as much
    	DEC stock as you can afford, financed by selling Microsoft short.
    	Sooner or later the market will agree with you as prices converge
    	to their true value (unless, of course, your perception is wrong).
    
    	best of luck, and may the FORCE be with you
        	/jim
    

�   	then, by all means, leverage yourself to the hilt and buy as much
�   	DEC stock as you can afford

Well, on Monday you can afford more DEC stock than ever before ... :-(

    Here's a little more polished news brief for our GREAT tpsA
    performance.  Spread the word!
    
    Mike
    
RDB SETS NEW TPC-A PERFORMANCE AND PRICE/PERFORMANCE RECORDS

On April 12, 1994, Digital announced Transaction Processing Performance Council
Benchmark A (TPC-A) results that set performance and price/performance records.
The TPC-A tests were run using client/server configurations running the ACMS
transaction processing  monitor and the DEC Rdb relational database software on
the OpenVMS operating system. 

A 4 node DEC 7000-650 AXP VMScluster achieved 3,692.02 tpsA (transactions
per second) at $4,873 per tpsA!  This is over three times the performance
of the previous record with a relational database, and even beats the
old non-relational database performance record set by IBM on the ES/9000
Model 511 running TPF. Each DEC 7000-650 node consisted of 5 Alpha AXP
processors, for a total of 20 processors.  In addition, the benchmark
configuration included 44 MicroVAX 3100 Model 90 client systems running
the ACMS transaction processing monitor to handle 36,960 terminals.

The 4 CPU Digital 2100 Server Model A500MP client/server system delivered
662.36 tpsA and set a new price/performance record of $4,401 per tpsA,
beating the record recently set by the DEC 4000-720 AXP client/server
system. 

These benchmarks show the capability of Alpha AXP Clusters, OpenVMS and
Rdb to handle extremely high throughput volumes against very large databases
while meeting stringent availability requirements.  It also clearly
demonstrates the scalable architectures of Alpha AXP, OpenVMS, and Rdb. 

Digital now has the four best TPC-A price/performance results, and the
result with the highest performance is one of the four!

Vendor    System                                        tpsA    $/tpsA
------------------------------------------------------------------------
Digital Digital 2100-A500MP 4 CPU C/S                   662.36  $4,401
Digital Digital 2100-A500MP 1 CPU C/S                   265.03  $4,405
Digital DEC 4000-720 AXP C/S                            402.76  $4,782
Digital 4 Node DEC 7000-650 AXP VMScluster C/S        3,692.02  $4,873

The TPC-A benchmark was designed to measure complete system
performance by simulating an on-line transaction processing (OLTP)
environment in which multiple users of terminals, workstations, or PCs
access and update a common database over local area networks.

Digital Equipment Corporation is the world's leader in open 
client/server solutions from personal computing to integrated 
worldwide information systems.  Digital's scalable Alpha AXP 
platforms, storage, networking, software and services, together with
industry-focused solutions from business partners, help organizations
compete and win in today's global marketplace. 

                               ####

Alpha AXP, AXP, DEC, Digital, the Digital logo, OpenVMS, and Rdb are
trademarks of Digital Equipment Corporation.

TPC Benchmark, TPC-A, and tpsA are trademarks of the Transaction
Processing Performance Council.

I've not seen any info on this on our WWW server... 

Lee

>I've not seen any info on this on our WWW server... 

The fact that it's NOT there indicates a lack of understanding and direction by
Digital's upper management...

Until upper management wakes up and understands that this announcement is GOOD
for Digital (hardware AND software) and NOT bad for Oracle, you won't see this
information published anywhere...

As Bob Palmer said at his meeting with Database Systems, Oracle runs on the
fastest hardware.  Well, this announcement certainly indicates that BOTH Digital
hardware and DEC Rdb are the fastest there is!

Rick

Rdb on OpenVMS/AXP shatters TPC-A world records

Performance World Record
========================
3692.02 tpsA @ $4873/tpsA on VMScluster of 4 DEC 7000 5-processor machine

Triples previous relational database record (also set on AXP)!

Even beats IBM Sabre system and IMS/Fast Path (non-relational) numbers!

This is mainframe performance at pc prices.

36000 users, 100 millions txns/day, 700 GB disk farm, 350 million table
rows, 8000 IOs/sec.

Price/Performance World Record
==============================
$4401/tpsA 662.36 tpsA on a 4-cpu DEC 2100-A500MP AXP

Beats even Compaq numbers on Intel 486, for example!

Scalable performance at constant record-low price/performance.

Top 4 price/performance spots for Rdb on OpenVMS/AXP.

System                                           tpsA   $/tpsA
--------------------------------------------------------------
Digital 2100-A500MP 4 CPU C/S                   662.36  $4,401
Digital 2100-A500MP 1 CPU C/S                   265.03  $4,405
DEC 4000-720 AXP C/S                            402.76  $4,782
4 Node DEC 7000-650 AXP VMScluster C/S        3,692.02  $4,873

    re -1 Rdb on OpenVMS/AXP shatters TPC-A world records
    
    SO WHAT !
    
    I've seen this message about 10 times in the last 2 weeks in notes
    files, mail etc. I've yet to see it published outside Digital. Until we
    tell someone in the outside world about it - AND they listen - it's just
    internal propaganda. I'm sick of us patting ourselves on the back about 
    what great products we have and then picking up the trade magazines and
    reading nothing about them. The newspapers in MA & NH may have a lot of
    Digital information but out here in the real world the press don't know
    a thing about us or our products ! 
    
    <flame_off>
    
    Tony.
    
    Ahhhh, feels good to get that off my chest.

    Re : -1

    Precisely our sentiment.  Digital is somehow reluctant to publicize
    this announcement to the rest of the world for fear of offending our
    partners!  What a nice company! :-)

    Ranga

    I just sent the RdB information to 400 DECUS members in Texas
    Oklahoma, and Arkansas...
    
    Published it in our LUG newsletter ...
    
    Sorry I don't hve budget for anything else...
    
    John Wisniewski
    
    (The American Airlines folks were flabbergasted that anyone could 
     even achieve more than 3500 TPS without using their IBM
     propriatary TFP?(TPF?) system;-)
    
     

Well let's see. I'm doing my part (but then again it's my job :^)

At German DECUS the guys here in Munich handed out over 100 copies of a fact
sheet and configuration of the benchmark to the attendees.

At DBWorld in London week before last I announced the results to more than 100
customers attending my talks, handed out 120 floppy disks with Rdb
presentations on them all of which contained the information and handed out
100+ color printouts of the hardware configuration with a fact sheet attached
(plagerized from Steve's memo :^). I also had a 2ft by 3ft color poster hanging
on the wall of the booth for all to see. (Phil Grice has the poster in his
office in Basingstoke).

At COMTEX in Moscow last week I announced the results to 150+ customers
attending the Digital presentations and handed out the other 100 floppies as
well. So we're trying.

If anyone wants the (very busy) slide of the hardware configuration I did along
with a slide for DBI you can get it from ZUSE::RDB$KIT:[SLIDES]DBWSIGN.PPT and
the fact sheet in TPC-3692.DOC (decwrite :^)

Pass them around!

Larry

>    I've seen this message about 10 times in the last 2 weeks in notes
>    files, mail etc. I've yet to see it published outside Digital. Until we
>    tell someone in the outside world about it - AND they listen - it's just
>    internal propaganda.

You've got the info - now go tell the world!  What's the problem???

A "grass roots" effort will have to suffice until someone in Marketing wakes up.

Rick

    does anyone here *know* anyone in marketing within DEC.  can we get
    some feedback to them directly.  I continually see that we need to
    market and that the 'non-marketing' part of the company knows it.  so
    how do we tell the 'other half'?

    
    FYI, if anyone would like copies, I think we finally have gotten most 
    of the typos out of the TPC Full Disclosure Report for this test; there
    are two files available as were reported to the TPC council:
    
    MSBCS::disk_walrath:[walrath.public]3k_summary.ps
    
    This is a two page summary of the configuration and 5-year pricing of the 
    3,692 TPS test.
    
    MSBCS::disk_walrath:[walrath.public]DEC_3K_TEST_TPCA.PS
    
    This is the "Full Disclosure Report" which gives a general description 
    of the configuration, and reports how all of the TPC-A requirements
    were met. It includes the above file, and is ~50 pages.
    
    When the final copy gets into TPSYS::SW_PERFORMANCE:[TPC.summary] and 
    TPSYS::SW_PERFORMANCE:[tpc.fdr] (these are the corperate repository for
    Rdb TPC results), I may delete these.
    
    Additional information (hopefully a nice technical report) should be 
    available in the not too distant furture.
    
    As a side note, I worked on both the Oracle/DEC 7660 1,079 TPS result,
    and on this test, and the amount of marketing ink spilled for the 
    Oracle results did seem to be substantially more than on this number.
    Pretty sad, IMO.
    
    Dave Walrath

    
    BTW, both of the files listed in the previous note are OK for external
    distribution. There may be a way to get bound copies of these for
    customers, but I can't remember how, and "VTX PERF" isn't available
    right now.
    
    DEW

RE: Note 2995.87 by MSBCS::WALRATH

>    As a side note, I worked on both the Oracle/DEC 7660 1,079 TPS result,
>    and on this test, and the amount of marketing ink spilled for the 
>    Oracle results did seem to be substantially more than on this number.
>    Pretty sad, IMO.

From what I have heard, out marketing people have been holding up on
marketing this great achievement.  It is possible that it will never be
marketed to any great amount...

This again shows me that marketing needs to be revamped greatly...

With these numbers we could be on front pages of MANY magazines with RDB
and Digital...

No wonder we are loosing money.

    re -1
    
    I hate to burst your bubble, but most of my customers don't really
    care.
    
    There are so many other issues that this is something that is nice to
    talk about internally, but will not sell any computers in my space.
    
    Ken
    A dumb 'ol field S/R

>    I hate to burst your bubble, but most of my customers don't really
>    care.

The customers *I* talk to DO care.  

Boeing (I'm sure you've heard of them :-) was astounded that such a TPS-rate was
even possible.  They were very enthusiastic about the "announcement".

Some of our other customers are requiring 10,000 TPS and waiting for the
database that can achieve this sustained rate...

Yes, some customers do not care.  But I think a fair number do care.

I think even more customers WOULD care if someone besides Oracle announced any
results!

Rick

    To add to the previous note:
    
    Customers would care if we recognized this achievement for what is
    means to their business and the industry in general and if we
    internally stopped this B**s**t 'doomsday' attitude!
    
    Folks, we can compete with this stuff if we take the proper steps in
    the sales qualification. Why is it most of our sales reps always start
    a DB sale opportunity with Oracle, Sybase or Informix? I've had ISVs
    tell me they are befuddled (we still use that word, right?) as to why,
    given the technical and business prowess of an Rdb, give up so
    sheepishly to Oracle and the like?
    
    Yes, they are our partners, we should offer their products if the
    customer explicitly says that is their strategic direction or it's a
    corporate standard/policy. Where the situation is entirely open for
    competition, and the sales person has qualified that this is in reality
    the case, there is no shame in leading with the most robust, powerful,
    performing, secure, manageable database in the industry. And price
    helps immeasurably in the selection.
    
    I'm not a DB expert, I work in Sales Support and have had reasonable
    success in promoting and winning Rdb sales, but it requires an
    attitude, tenaciousness, and a willingness to invest time to helping
    the client become aware, and even understand, why their business needs
    these capabilities.
    
    regards,
    Ruben in Southern Cal.

Greetings!

  Has anybody here (or in Rdb) actually contacted someone who can disseminate
this rather remarkble achievement to the customers.  I hate to state the ob-
vous but a bunch of DECies in this notesfile are all preaching to the choir.

  I just checked Digital's "Marketing home page" on our Web server under 
"new annoucements", "press releases", "performance informance" and just about
anywhere else I could a customer might look and saw nothing about the Rdb
record.  I'm on some sort of distribution list for "press releases" and haven't
seen anything,  I just got the latest issue of "Customer Sales Update" and did
not see anything (perhaps the printer's cutoff date was too close).

  All this is very sad in my opinion, but then the folks who update the Web in-
formation may simply not be aware.  The person's name that comes up the most 
often relating to the Web home page is Russ Jones.  I'm going to forward .0 off 
to him (as soon as I get permission from the author) and see if Russ is even 
aware of it, perhaps someone more qualified would like to "dress up" .0 a bit 
and make a snazzy page (or two) under appropriate page on our Web server, but
keeping this achievement a "secret" just doesn't make sense to me.  No try, just
do, or do not...(apologies to Yoda ;)...)

						Tony Swierkowski
						SDCC (in Palo Alto)
						(415) 617-3601
						"[email protected]"

    In April, a decision was made to not issue the press release about
    the Rdb TPC numbers talked about in this notes stream.  With the
    recent change of management, etc, this decision has been reversed
    and the press release will be issued this week.  If you have specific
    questions on this, contact Mike O'Connell, the DEC Rdb marketing 
    manager.
    
    regards,
    Andy Schneider
    DEC Rdb for OpenVMS Product Manager
    

    .94
    
    Are you hinting that Lucente was the one who opposed publishing the Rdb 
    TPC numbers ?

Re: .93

It is in the world wide web but it is hard to find. We (John Apps really :^) 
found it under System Performance and somewhere else I believe. I'll have to go
have another look. But Russ or whoever needs to put it with the results it
belongs with. 

Re: .96 
	:^)

    Why on *earth* would anyone want to keep this information secret? I
    must be stupid or something, but I see no benefit in that at all...
    
    Laurie.

    We were "soft-pedaling" these benchmark data (one assumes) because,
    were they have been promulgated as widely as imho they deserve to be,
    they would presumably have ruffled the feathers of our key partners in
    the database space, e.g., Oracle.  At the time that this assumption was
    in force, we must have preferred to do our competing from the supine
    position, I can only surmise.
    
    /s/Dan--whose_elder_daughter_is_a_Product_Manager_@_Oracle!
    

     Our world records are not a secret any more. Digital has now come out
     with a press release on our record-breaking numbers.  This should put
     to rest some of the recent claims by SGI that they have the world's
     best numbers (with only 2K tps!). Efforts are under way to do ads and
     to carry the news in the media. It is great to have your company help
     you a little! :-)

     Ranga
--------------------------------------------------------------------------------
     David Farmer
     (508) 493-0179





         DIGITAL BREAKS WORLD RECORD FOR RELATIONAL DATABASE
                    PERFORMANCE, PRICE/PERFORMANCE


         ... Nearly doubles previous performance records ...



MAYNARD, Mass. -- May 5, 1994 -- Digital Equipment Corporation 
announced today record-breaking performance and price/performance 
Transaction Processing Performance Council Benchmark A (TPC-A) 
results. The results represent significant advancements over vendors 
such as Silicon Graphics, IBM, Hewlett-Packard, SUN, Sequent and 
COMPAQ. 
     TPC-A tests were run using client/server configurations running 
Digital's ACMS transaction processing  monitor and the DEC Rdb 
relational database software on the OpenVMS operating system.
     A four-node DEC 7000-650 AXP VMScluster achieved 3,692.02 tpsA 
(transactions per second) at $4,873 per tpsA.  The new performance 
record exceeds all previous TPC-A results by any relational database 
by five-fold on Hewlett-Packard and SUN platforms, and by three-fold 
on Sequent platforms. It also beats the non-relational TPF  system 
(a highly specialized transaction processing environment) running on 
an IBM mainframe, and achieves lower price per tpsA than IBM and 
COMPAQ systems. This test also demonstrated, for the first time, 
support for over 36,000 concurrent users, processing over 100 
million transactions in an 8-hour day for less than $5000 per tpsA.

     Each DEC 7000-650 node consisted of five Alpha AXP processors, 
for a total of 20 processors.  In addition, the benchmark 
configuration included 44 MicroVAX 3100 Model 90 client systems 
running the ACMS transaction processing monitor to handle 36,960 
terminals.
     These results reinforce the long-standing strength of Digital's 
cluster technology. In this case a DEC Rdb database larger than 700 
gigabytes, with one of the tables containing over 350 million rows, 
provided shared access from all nodes and processors in the cluster. 
The system executed over 8000 disk I/Os per second. 
     Also part of the test series, a four-CPU Digital 2100 Server 
Model A500MP client/server system delivered 662.36 tpsA and set a 
new price/performance record of $4,401 per tpsA. The results beat 
the record recently set by the DEC 4000-720 AXP client/server 
system. 
     "These benchmarks show the capability of Alpha AXP clusters, 
the OpenVMS operating system and DEC Rdb to handle extremely large 
throughput volumes against very large databases, while meeting 
stringent availability requirements," said Rose Ann Giordano, vice 
president, Production Systems Software. "Digital continues to 
demonstrate its ability to provide a powerful environment for high 
performance, high availability, and concurrent access to massive 
amounts of data, all critical elements of today's application 
downsizing efforts."
     Digital now has the four best TPC-A price/performance results, 
and the result with the highest performance. They include:
    
    System                                    tpsA     $/tpsA
    
    Digital 2100 Server Model A500MP 
    (4 CPU C/S)                   	      662.36   $4,401
    
    Digital 2100 Server Model A500MP 
    (1 CPU C/S)           		      265.03   4,405
    
    DEC 4000-720 AXP C/S                      402.76   4,782
    
    4-node DEC 7000-650 AXP VMScluster C/S  3,692.02   4,873
    

     The TPC-A benchmark was designed to measure complete system 
performance by simulating an on-line transaction processing (OLTP) 
environment in which multiple users of terminals, workstations, or 
personal computers access and update a common database over local 
area networks.
     Digital Equipment Corporation is the world's leader in open 
client/server solutions from personal computing to integrated 
worldwide information systems.  Digital's scalable Alpha AXP 
platforms, storage, networking, software and services, together with 
industry-focused solutions from business partners, help 
organizations compete and win in today's global marketplace. 
                               ####

Note to Editors:  Alpha AXP, AXP, DEC, Digital, the Digital logo, 
                  OpenVMS, and DEC Rdb are trademarks of Digital 
                  Equipment Corporation.

     		  TPC Benchmark, TPC-A, and tpsA are trademarks of 
                  the Transaction Processing Performance Council.

CORP/94/476

    
	Looks like Oracle is fighting back with 'best' numbers for 
        'Open Systems' angle. They do have some legitimacy for customers
        (a majority) looking for Unix/WNT solutions. So, till Rdb can
        run on OSF/1 and produce similar numbers, Oracle will still be
        the fastest option for most customers.
    
    -------------------------------------------------------------------------
    
    
    
    
         News Article: ORACLE ACHIEVES RECORD PERFORMANCE ON IBM RISC
                       SYSTEM/6000 &
 
         25-MAY-1994   Length: 92   lines           Businesswire

   REDWOOD SHORES, Calif.--(BUSINESS WIRE)--May 24, 1994--Oracle Corp. 
announced today that it has attained the highest TPC-A benchmark 
performance ever recorded for a commercial, uniprocessor database system 
in the open systems market and one of the highest benchmark results 
for a loosely coupled cluster.
                                                                             
	   In a TPC-A benchmark, the Oracle7 database running on a single-node 
IBM RISC System/6000 Model R24 achieved 357 transactions per second 
(TPS-A) at a price performance of $7,334 per transaction ($/TPS-A).  On 
a 4-node HACMP/6000 cluster, Oracle7 with the Oracle Parallel Server 
achieved a TPC-A benchmark result of 894 transactions (TPS-A) at a 
price performance of $8,461/TPS-A.  See chart below for details.

	   The Model R24 from IBM uses IBM's new POWER2 multi-chip 
microprocessor and offers the highest performance in the RISC 
System/6000 family of servers.

	   The Oracle Parallel Server enables loosely coupled systems, such 
as IBM's HACMP/6000, to achieve high availability and high 
scalability.  For customers, OPS means the ability to add RS/6000s to a 
cluster as needed and to provide 7x24 applications availability at an 
affordable price.

	   "With these results, we have demonstrated how our Parallel Server 
technology best exploits the RS/6000 system," said James C. Sha, vice 
president, UNIX products division, Oracle.  "The combination of Oracle7 
and the IBM R24 cluster is an excellent solution for customers looking 
for open, high-end database servers."

	   "The record set by this benchmark is good news for RISC 
System/6000 customers," said Dave Cassano, vice president, Systems 
Marketing, IBM RISC System/6000 Division.  "The Model R24 is the 
highest performing IBM UNIX uniprocessor system in the marketplace 
today, and IBM is excited to be able to offer this level of performance 
and high availability cluster capability as part of the RISC 
System/6000."

	   The Oracle and IBM relationship has been a productive one for both 
companies.  The close development relationship allows Oracle and IBM to 
continue to deliver highly tuned, optimized solutions to the market 
first, and allows Oracle and IBM marketing teams to work cooperatively 
with customers deploying Oracle solutions across their enterprises.
-0-
*T

SYSTEM      RDBMS     OPERATING    NUMBER      TPS-A    $/TPS
                      SYSTEM       OF NODES

RS/6000     Oracle7   AIX 3.2.5    1           357      $7,334
Model R24

RS/6000     Oracle7   AIX 3.2.5    4           894      $8,461
Model R24

*T
-0-


	   The IBM RISC System/6000 Division based in Austin is responsible 
for the design, development, and manufacture of the RISC System/6000 
family of servers, the AIX/6000 operating system, and the PowerPC and 
POWER2 microprocessors.  Systems based on these industry-leading 
microprocessor designs support the more than 10,000 software 
applications available for IBM's AIX/6000 operating system, enabling 
users to maximize the power of the RISC System/6000 to meet their 
business needs.  The IBM RISC System/6000 Division is committed to 
providing commercial and technical customers with best-of-breed, 
RISC-based hardware and software solutions, which support open system 
standards.  The RISC System/6000 product line is marketed in the United 
States and internationally through the IBM sales force and IBM Business 
Partners.
	
  Oracle Corp., with headquarters in Redwood Shores, Calif., is a 
leading supplier of information management software.  Oracle develops
and markets the Oracle7 family of software products for database 
management; Cooperative Development Environment (CDE), a complete set
of CASE and application development tools for enterprise-wide, 
client/server computing; and Oracle Applications, packaged 
client/server solutions for human resources, accounting and 
manufacturing.  Oracle software runs on personal digital assistants, 
PCs, workstations, minicomputers, mainframes and massively parallel 
computers.  The company offers its products, along with related 
consulting, education and support services, in more than 90 countries
around the world.  

	   -0-

	   Note to Editors:  Oracle is a registered trademark and Oracle7, 
Cooperative Development Environment and Oracle Parallel Server are 
trademarks of Oracle Corp.  Other product or company names may be 
trademarks of their respective holders.

    They won't be able to do this much longer as they have achieved their
    numbers using a benchmark special. It has been said that their
    performance would be in the Oracle V6 range if they didn't use the
    benchmark special.
    
    michael

	re: "Benchmark special"

	As Mike pointed out, Oracle gets their performance numbers by using
	a "feature" called discrete transactions. This is a feature that, as
	I understand it, is useless in a "real" environment. This is being
	disallowed shortly, which is probably a major reason why Oracle has
	decided to stop doing TPC benchmarks. Time will tell whether they
	can drive the standards, or whether the standards will survive well
	without them. (Look how well the Microchannel bus worked for IBM in
	the PC market -- NOT).

	re: Numbers

	It's still interesting to look at the numbers:

DBMS	        System                                  TPS A     $/TPS A

DEC Rdb	 Digital 2100-A500MP 4 CPU Client/Server        662.32   $4,401
DEC Rdb	 Digital 2100-A500MP 1 CPU Client/Server        265.03   $4,405
DEC Rdb	 Digital DEC 4000-720 AXP 2 CPU C/S		402.76	 $4,861
DEC Rdb	 Digital DEC 7650 AXP 4 nodes		       3692.67	 $4,866

Oracle7	 RS/6000/AIX 3.2.5    1 CPU           		357      $7,334
Oracle7	 RS/6000/AIX 3.2.5    4 CPUs          		894      $8,461

	Yes, their numbers on the RS6000 beat ours on the 2100 straight-up.
	But look at the price-performance -- they cost nearly twice as much!
	So, if you're interested in a system that really cranks out the
	performance, the Digital solution still gives the best bang for the
	buck. And Oracle hasn't approached our high-end numbers. At least
	not yet.

	Of course we'll still get killed in the marketplace by being
	labelled as "closed" and "proprietary," at least until we get some
	OSF/1 benchmarking done.

	Roy

    Wasn't OpenVMS XPG branded? Isn't OpenVMS POSIX compliant? COSE and CDE
    aren't that far behind!!
    
    Elevate the "open" question to one of industry standards compliance
    (unless the customer explicitly states that it's UNIX of course). DEC
    OSF/1 has many features that gives you advanteges. Lead with them...
    
    
    regards,
    Ruben

>                          -< But OpenVMS is open... >-
>
>    Wasn't OpenVMS XPG branded? Isn't OpenVMS POSIX compliant? COSE and CDE
>    aren't that far behind!!

	An ISO number isn't what the customer means by "open" - if you
	choose an "OpenVMS" solution you can only have a single source
	for processor-based hardware ... that ain't "open." The people
	I've dealt with (utility, software, healthcare, shipping, military,
	communications, entertainment) don't care about what x.y.z branding
	we have or will have in the future. Don't get me wrong, I love VMS
	but it ain't open in any way the people who make the "let's buy
	it" choices seem to care about.

	Larry_out_with_the_common_folk_who_aren't_on_any_standards_groups

RE: .105

	Then, praytell, what defines "Open" to these people? 
	Incompatible Unix systems?

							mike

Re: .106

>RE: .105
>
>	Then, praytell, what defines "Open" to these people? 
>	Incompatible Unix systems?
>
>							mike

	BINGO!!!

Give this man a cigar!

    Let's see:
    
    HP is the only source for HP hardware
    SUN is the only source for SUN hardware (they've effectively killed
    their clones)
    IBM is the only source for IBM hardware...
    
    Yeah... I think I'm getting the hang of this open thing...
    
    regards,
    Ruben

"Open" also has to do with the Applications.
I worked with one customer who defined an "Open Server Architecture".
They had a list of approved vendors of computer systems.

Any new applications developed at the corporation, regardless of which
platform they were implemented on, had to pass their test suites on 
ALL approved platforms.

That means that the customer has the ability to switch vendors whenever
the price landscape changes, with minimal disruption to ongoing programs.

It makes it a much more brutal selling environment.

Kevin Farlee

>    HP is the only source for HP hardware
>    SUN is the only source for SUN hardware (they've effectively killed
>    their clones)
>    IBM is the only source for IBM hardware...

    The customers weren't buying from these folks either ... since we're
    talking about transactions per second I think it worth noting that
    in the past ten years of working on OLTP applications I only had
    two opportunities (out of a hundred or so) that needed more than
    10-15 TPS, and the datacenter folks were more interested in finding
    a simple & easy solution than they were in transaction rates of
    benchmarks; heritical I know, but that's reality - leading with a
    TPC benchmark is leading with your chin.

    The _only_ thing that closes a production systems sale is if the vendor
    shows they understand the customer business and can demonstrate a
    credible solution that the customer can purchase or build/maintain
    at a reasonable cost - you can't sell TP systems via 1-800-BUYME and
    a couple of glossy SPD's.

    As for "open" I agree with your comments, but if the customer thinks
    that open = unix then we better start getting the snake oil into our
    bottle ... saying that "they don't know what they want" shows an
    arrogance we can't afford; they might not know what they want, but
    they have a good grasp on what they don't want and the only place
    you can sell VMS today is into the installed base (in my experience).

    That turnip is almost out of blood.

   Larry

OPEN Software has several dimensions:		Is Rdb open?

Compliant with industry standards	  	SQL92, XPG4?
Compliant with defacto standards		ODBC?
Runs on UNIX					OSF1?
Runs on Several O/S 				NT,OSF,VMS?
Runs on Several Hardware Platforms 		Alpha, VAX, Intel?
Runs on multiple hardware vendors   		Digital, HP, Sun
Software built by a software only company, + above.

Vendor who does the MOST ADVERTISING!

Therefore;

Is IBM's DB2 (IBM + HP) OPEN?
Is CICS (IBM, Alpha, ...) OPEN?

or

If you are BIG enough, does anyone care? (defacto open)


OPEN is how you define it and how you sell it. No, VMS isn't 'OPEN' really but
then neither are the other proprietary O/S of other companies from which they
also make money like we do. But we DO have OSF1 and NT which are 'OPEN' as
viewed by our customers even though they aren't either. OSF1 is UNIX but is any
UNIX really the same as another yet, or stable enough and feature rich enough
to really be a production system? And NT, comes from ONE software manufacturer
who (when you get right down to it) wants to dominate the world, is that OPEN?

Rdb is working it's way towards 'OPEN' but again based on the list above we are
to some extent, just need to get better. The real kicker is that while ORACLE
sells itself as the 'OPEN' solution you as a customer are completely and
irreversibly locked into ORACLE. Is this 'OPEN'? What we're trying to do in
Database Systems here in Digital is to provide the best world class industry
strength (marketing :^) database system that runs on several platforms pairs
with more in the future while at the same time providing Data Integration tools
that truly make a multi database environment 'OPEN' in the sense that your
applications will run ANYWHERE on ANY DATABASE.

By the way, Rdb will completely cover the first 5 items in the list above
within the next 12 months, this is NOT vaporware.

Larry

    The definitions of "Open" I have tried on various people including
    customers and received no disagreement is:
    
    "An OPEN product/system is available from multiple vendors in price
    competition with each other with a reasonable degree of substitution
    between vendor's products".
    
    It's a marketplace definition, it has nothing to do with technology.
    MS/DOS, although developed by Microsoft, is available throuh multiple
    sources. UNIX applications can have a reasonable degree of portability
    between different flavours of UNIX. Most customers want to be able to
    buy the latest hot application on the latest hot box without having to
    make any commitment to any one vendor. Certificates of "openness" from
    standards authorities in most cases are little better than wallpaper,
    unless they really indicate substitutability between products from
    multiple vendors.

    re:.-1
    Exacty.  But I think another good definition of "open", as used by the
    only folks who count (customers), is simply a modification of a famous
    judge's (Blackmun?) definition of pornography:
    
    	I can't tell you what it is, but I know it when I see it.