Conference ulysse::rdb_vms_competition

>1) Sybase was cheaper  $400,000 CDN vs. $670,000 CDN for one development 
>    		       licence, four interactive licences and all the tools
>     		       (VAXset,TEAMDATA, RALLY, CDD/Plus).  
>    		       They even left their complete pricebook with the 
>    		       customer.
 
    VAXset ? How were they going to use this with the above tools ? VAXset
    + Rally are VERY expensive, Rdb is cheap, are you sure you were
    comparing like with like ?
    What about Rdb's free run time licence or the interactive license.
    
>2) Sybase will support SMP in June and has it running at several referenceable
>                       BETA test sites.  In fact Sybase recommended an SMP 
>    		       platform for this application.
    
    HOW is their client single server architecture going to SUPPORT SMP. 
    How easy will their implementation be to manage ?

>3) Sybase supports VAXclusters.  This architecture is three VAX 6000's in a 
>                       C.I. cluster and Sybase assured the customer it would 
>                       run fine.
    
    INCORRECT. Have they suddenly started to use DLM ? Yes Sybase may run
    in a cluster BUT can they access ONE instance of a database
    simultaneously from 2 or more nodes? The answer is NO. 

>4) SQL services for PC's is one of the major advantages that Sybase won over 
>                       RDB 
    
    What ? Sql Services is OUR product that works to OUR advantage.
       
>5) SYBASE has 2 PC    

    WE support 2PC, not just over Rdb but over Rdb, DBMS + RMS files.
    
>6) In the technical evaluation Sybase was the clear leader over RDB, Ingres, 
>                       and Oracle.

    In whose technical evaluation ? I, along with alot of Rdb people, do 
    not believe this. 

>What I would like to know who, if anyone, has this information or is able to 
>confirm it?  We have made it quite clear that if SYBASE is not up to scratch, 
>we will not take the blame with our architecture, as we had warned them with 
>what information we had.

    I strongly recommend that you get hold of Mike Booths competitive fact
    sheet for Sybase. It is a great pity that you had not read this
    document BEFORE you tried to sell Rdb. Did you use the standard Rdb
    slide set for your presentations ? 
     
>Unfortunately, I feel a little silly, as I had no clue as to the above 
>information (except SQL), I just hope this will help someone else.

    Sigh.
    
I will also but this in the Competitive notes file.


    
    

    I have read the stuff - I am not totally in my shell up here.
    
    I almost quoted it verbatim - but SYBASE assured them that this
    beta test site was referenceable, and that this was all new.
    
    While I have no problems that we did not compare apples to apples
    - re CDD/Plus etc.  They actually already have it (CDD/Plus) on
    another system, but they wanted a fully/tightly coupled package
    in this case.  Previous to this they had only worked with RDB V2
    ish - so we also had to convince them how much it had changed and
    how etc.   
    
    They require the interactive licence as they need the SQL for
    interactive queries, and they want to manipulate the DB definitions.
    YES - all want to do this - not just one node.  It was all we could
    do to let them have only one FD licence, as they are programmers
    at heart.
    
    They also did not want to reinvent the wheel when it comes to SQL
    on a PC
    - so they went with stuff already available that talks to SYBASE.
    At present we have very little to offer unless they write their
    own applications, which they explictly did not want to do.
    
    The VAXcluster issue per se is a non-issue re fail-over server
    that SYBASE offers, as each node is doing its own thing, with no
    real sharing of resources.  Why they are on a cluster, I believe,
    is that a single real time DB only needs to pass information only
    once, each node will do it's own thing with that DB info. 
    
                    
    We do not have 2PC out yet - and cannot deliver it now - so even
    though it was stressed - it will be a part of the operating system
    etc. - it didn't make a difference.
    
    I don't know why I am trying to defend their selection - we came
    in at the very last minute - less than 4 days to prepare an RFI
    and that was that.  Before you shout DEMO - we had no time at all to demo 
    anything. 
    
    I was not around for the PM on why we lost so I am only reiterating
    what was told to the salesman concerned.

    
APT Data Services   COMPUTER INDUSTRY  -  NEWS SERVICE   [22-MAR-90]
London     							

SYBASE SQL SERVER ADDS SYMMETRIC MULTIPROCESSING SUPPORT

Sybase Inc, Emeryville, California yesterday announced Sybase
SQL Server support for symmetric multiprocessor systems, saying
that it had extended its uniprocessor server architecture to a
Virtual Server Architecture explicitly designed to take full
advantage of symmetric multiprocessor hardware systems to
provide high throughput and linear scalable performance. The SQL
Server with Virtual Server Architecture will be available for
beta testing on the Stratus XA2000 line of fault-tolerant
computers next quarter and generally available later this year.
Versions will follow for DEC and Pyramid Technology computers
later in the year. The new version manages all database
scheduling, load balancing and synchronisation without
interference from the operating system. On a Stratus XA2000
Model 160 in a fault-tolerant environment with of networked
communication and duplexed disks, the thing did 43 TP1
transactions per sec ond with an average response time of 0.71
seconds on a database with over 4.3m rows - against 8.1 tps with
the uniprocessor version of SQL Server, Sybase says. Scalabil
ity is claimed to be 95% when the number of processors goes from
one to six. Memory needed is 50Kb per user. The company gave no
prices.
    

    If they haven't signed the contract with Sybase, I strongly suggest
    you get some of our Canadian resources who TRULY understand databases,
    enough to seperate fact from fiction. The customer it seems has
    bought into all the crap from Sybase. Does the customer understand
    issues such as integrity and timeout problems with Sybase -the concepts
    of Checkpoints if explained correctly to the customer should scare
    them sh*itless. There are other things: Multifile database, row-level
    locking, referential integrity with stored procedure. These can
    be severe headaches!
    
    What about support? I'm sure they have absolutely no-one in Alberta.
    These reference sites you talk about, where are they in Canada?
    I believe Alberta Cancer Inst. is one of the few in the Calgary
    area.
    
    The customer needs to hear the truth both about Sybase and Rdb.
    If its not too late I strongly suggest you call Ken McMullen @TRC
    or Janet Spackman @TRO

    Sybase does two phase commit?  Sure they do, but here's the difference. 
    They do it for *every* transaction, and only to a single node at once. 
    Not a distributed update to two or more nodes, which is why you want a
    2pc protocol, after all.
    
    Sybase's ability to do 2pc can be picked apart very easily.  It buys
    the customer nothing and contributes additional overhead.  (That's why
    we put a TRUE 2pc protocol into VMS, and not the database.  
    
    By the way,  I would venture to guess that the fact that 2pc is a few
    months out should have little or no effect upon how they code their
    application.  The fact that a transaction occurs and that a 2pc
    protocol enhances the ability of the transaction  to complete properly
    should be transparent to the database (once again...with Rdb, the 2pc
    occurs inside the OS/Network levels..)
    
    Sybase SQL - very proprietary.  Takes extra code inside the application
    to talk to the Sybase Server - NON ANSI SQL code.
    
    LSE support.  I don;t believe Sybase has any kind of language sensitive
    editor for SQL code generation.
    
    Dectrace/rdbexpert - What kinds of tools are available to help tune the
    database?  Most of these third party databases have little, if any,
    tools to tune the database in any specific environment.  They are a
    BLACK BOX to the DBA in terms of tuneability.
    
    Running on a cluster?  Sybase convinces the customer that they really
    don't need to have the db engine on multiple nodes, because sybase
    can't do it.  It goes against their server strategy.  They point
    fingers at the DLM as being slower, because sybase has nothing like it. 
    This is the discussion of just running on a cluster versus taking
    ADVANTAGE of the clusters capabilities for providing a higher level of
    failure tolerance.
    
    Have the customer take a program (pick your 3gl) with imbedded SQL
    written to talk to a sybase database.  Then convert it to VAX SQL and
    examine the difference in lines of code it takes.  
    
    Reference sites have limited value if the reference site has limited
    experience.
    
    Hope this helps!
    
    Gil
     
    
    
    

    
    I did realize that what they say and what they really do are two
    different things.  Technically Sybase functions on clusters and
    SMP but does not take advantage of these.  This apparently was not
    a big deal to the customer,  which is why we made very sure that
    they understood this.  We were only able to talk to the Project
    Leader - who is very PRO Rdb - but unfortunately there were about
    14 people making the selection - and we had no access to them. 
    I also realize that Sybase SQL is non-standard, but if you cannot
    talk to those who make the decisions, you cannot belabour the point.
    
    I have subsequently found out that they already had a strong investment
    elsewhere in SYBASE and that this information was not released to
    us until after the decision was made.
    
    I appreciate what you have pointed out to me, unfortunately except
    for Sybase's 2 phase commit, I had everything just about covered.