Conference bulova::decw_jan-89_to_nov-90

    (1) For DECwindows, roughly equal                         
    
    (2) Is this a DECwindows question? This is a matter of licensing,
    first, and application mix second.
    
    (3) The short answer is "yes".  If your concealed question is "When
    will Digital drop support for VMS?"  The short answer is "See Figure
    1". The somewhat longer answer is "Not likely in my lifetime".
    
    (4) Ditto, yes.
    
    The thematic answer is that over time, availabiliy of applications 
    for a platform will determine the choice of that platform.

Since the question is in the DECwindows conference, I assume that this has
something to do with DECwindows.  When I use DECwindows on VMS and Ultrix
and on both and when using MIT X11R3, I see no difference in performance.
This is intended.  Ideally the issue of which is better will GO AWAY.

Neither VMS nor Ultrix makes a difference in the number of interactive users;
in both cases, one is all that is comfortable per workstation.  As they say,
two's company and three's a crowd.

I'm not sure about the customer's investment, but either one is good for my
investment in DEC stock.  Preferably, the investment should be in a nice
expensive workstation, how about one with 3D graphics and multiple CPUs.

Seriously, I don't mean to offend, but these questions should not be asked
in ANY conference.  It makes as much sense as asking which is better, breaking
the egg on the big end or on the little end.  That, by the way, was a pointed
way to saying that the issues of religion in England didn't make any sense.
In other words, this is a religious issue, and any answer you get here will
matter only if the customer you pass it onto is of the same religious belief.

    re .1,.2:  Aren't we over-reacting a bit?  As a disinterested third
    party, taking .0 at face value, I see no hidden agenda (unless I'm
    naive, which could well be the case).  Giving straight answers to
    innocent questions (which I suspect came from some customer) is
    probably all that is required.
    
  --Simon
    

I agree with .3, this is a perfectly sensible question and of course there are
real differences. One of the goals of DECwindows is to try to make the difference
invisible to users and application programmers (UIL/XUI Toolkit only) but
that does not mean that there are'nt differences. When choosing which type of
system then these differences have to be explained. 

If its as simple as "you want a 14 MIPS workstation" then (today) you better 
think of ULTRIX... 

Within Digital it seems that we are afraid to talk about the differences that
exist between UN*X and VMS. Clearly there is the overall question of differences
and one specific to the workstation environment which should be OK to discuss
in this conference... 

If a customer asks a specialist or salesman in the field - what benfitis and
what would you recommend for my environment ULTRIX or VMS for my DECwindows 
workstations are we really going to say...

> It makes as much sense as asking which is better, breaking
>the egg on the big end or on the little end.  That, by the way, was a pointed
>way to saying that the issues of religion in England didn't make any sense.
>In other words, this is a religious issue, and any answer you get here will
>matter only if the customer you pass it onto is of the same religious belief.


There are technical difference that get overwhelmed by personal beliefs and it 
is this that makes it seem like a religious subject.

He is my technical evaluation (from the VMS perspective...)

Rememebr that this in not from the users perspective (as such)...

Given that the performance, facilities and appearance are similar, the choice
comes down the fundamental operating system qualities. It is 
generally regarded that VMS is the leading vendor specific operating system
and that Ultrix is the best portable operating system. So the general
guidelines should be :

1/ If its quality & flexibility (i.e. commercial environment) then go VMS

2/ if vendor independence is paramount or they just plan only like UNIX
   then Ultrix is the better choice.

VMS however has a number of benefits over Ultrix, which must be taken
into consideration.

VMS offers flexibility in the form of VAXclusters. This is important when
considering large installations of workstations which will benefit greatly
from Local Area VAXclusters (LAVc).

VMS offers extensive System Management facilities, again this is very 
important in highly populated workstation environments where central 
system management is essential.

Because of the richness, coheriency and sheer longevity of VMS, there exists
an extensive set of Layered Products. Although DECwindows has the
capability of having a distributed environment is predicted that 
(at least initially) the predominant use will be local application usage. In
this case the extensive set of layered products under VMS will be attractive
to customers and to Digital in terms of software license revenue.

Ultrix does not have the concept of shareable libraries. One of 
the results of this is that each programme, particularly if it is a DECwindows
application, tends to require excessive disk space. This is important
in keeping the cost of each workstation down. Furthermore VMS has the
concept of shareable files, which means that in a VAXstation LAVc only
one copy of VMS is needed, thus saving considerable amounts of disk space and
possibly memory.
By way of example, the OOTB applications take consume between
4,000 & 8,000 blocks each, (on a DECstation 3100) that can soon eat up an 
RZ23/RDZ54.

Desktop VMS means that installing and maintaining a group of work stations
by non-technical people is a lot easier. Such facilities do not exist for ULTRIX.

These are important factors, particularly in the commercial markets, when
choosing a computer system, irrespective of whether or not it is a windowing
environment.

It is important that Digital leverage hardware and software sales 
from the VMS implementation of DECwindows. In particular, the 
LAVc configurations where the diskless systems and consolidated 
system management give Digital an advantage over other workstation 
manufactures. 


  

re: .3, .0

I find it refreshing to hear a VMS engineer give a reasonably fair (if
slightly biased) answer to this question.  I'd really like to see more of 
this.  If we could get a good honest discussion going here, like in the SCS 
note, I think all would benefit.  When I was back in marketing, no one would 
touch the real guts of the issue with a ten foot pole.  I'd *LOVE* to see 
the ULTRIX and VMS internals people go head to head with some honest
discussion about this.  

I'm just an application-programmer-turned-marketing-twinkie, but I did code
under VMS and UNIX for nine and five years respectively(still do), so I'll
give it a shot from a slightly different angle.... 

I agree that XUI is the canniest move we could have made to smooth over the 
differences, but let's look at the rest of the picture.  I'll summarize
before I start, so you don't have to read all my reasons and ramblings, if
you don't want to. 

  1.  If your application only runs on one or the other, pick the place it runs.
  2.  If you need some of the commercial, high-quality features of VMS, go VMS.
  3.  If you need to merge SEAMLESSLY into a heterogeneous environment,
        (timesharing or workstation) choose ULTRIX.
  4.  For a programmer, choose the o/s philosophy that best fits your style.

I'd also like to add that just because VMS or ULTRIX doesn't have a piece of 
functionality now, that doesn't mean it always won't.  VMS *IS* more mature 
than ULTRIX - we've been coding in-house on VMS lots longer, so have more of 
the "careful planning" KO loves to talk about.  VMS has done some great
stuff that UNIX people would love to have, but don't have YET(SMP, for 
example).  ULTRIX has some stuff that VMS would love to have but don't have
YET(POSIX, for example). 

APPLICATIONS

Most workstation sales are now made on the basis of application 
availability.  If the application only runs on one of the platforms(which 
will become less likely with DECwindows), choose that platform.  I'd like to 
point out that the majority of workstations applications are written for 
UNIX.  At BBN, other than the graphics, it took six weeks to port between 
UNIX platfors.  It took 18 months to port bto new proprietary ones.

FUNCTIONALITY

If things such as security, vol shad, SMP, shareable libraries/programs/images 
(including re-entrant shareable images), direct memory mapping, etc are GOT
TO HAVE TODAY criteria, you'd be a fool not to choose VMS.  I've attached a
chart that comes from a UNIX to UNIX to VMS comparison chart being done.  If 
I had left the other incomplete columns, you'd see that for the functionality
listed, VMS wins HANS DOWN.

**BUT**

What will OSF add to this equation?  Hard to say.  It may help ULTRIX catch 
up quickly, or it may be a millstone.  Only time will tell.

HETEROGENEITY

VMS will not plug and play in a heterogeneous environment as easily at 
ULTRIX.  When you take ULTRIX "out of the box", it has the guts of what is 
needed to merge with other workstation vendors' products: NFS, TCP/IP, pcc,
common tools(make, yacc, awk, vi, sccs, etc) and X11 (either DECwindows or 
pure X11R{2,3}<-(depending).  VMS starts off with none of those.  Yes, you 
can get DEC/Shell and UCX and be PARTWAY there, but you have to BUY them and
they still don't get you completely merged.

STYLE

I teethed on VMS, but I prefer UNIX.  Why?  I like pipes, fork/exec, tons 
of little utilities, of which *I* am the master.  I can easily create 
new tools from a bunch of existing ones - they plug and play well together.  
Once I found more awk, and the grep family, I was hooked.  Not so everyone.

On the other hand, I know a very well respected programmer who was driven 
BONKERS by all the inconsistencies of UNIX.  He didn't CARE that you had all 
these little toys.  He wanted CONSISTENT QUALIFIERS FOR EVERY COMMAND.

This is the religion part.  Here, you let people test drive it to find out.  
As with cars, some like sports cars and some like luxury cars and only the 
foolish salesperson tries to change their mind.

Here is the chart I mentioned.  This was a quick and dirty to get a project 
going(ask SSGBPM where the finished one is) before I left.  IT IS 
INCOMPLETE, OUT OF DATE AND MAY CONTAIN SOME ERRORS.  


          Functionality Comparison among UNIX Variants and VMS

                        ***FIRST DRAFT***






Feature          Ultrix        VMS      System V.3     SunOS      HP      IBM
POSIX              X            O          O             O        O        
X/Open             X            O          X             O        X
OSF                X            O          O             O        X
X-Windows          X            X          O             PART     X
SMP                O            X          X             Planned  O
Distrib. File Sys  X            DFS        X             X        X
    NFS            X            PART       O             X        X
    RFS            O            O          X             ?        ?
Security(C2)       O            X          X          B1 Coming   ?
Vol Shadowing      O            X          X             X        ?
(server failover)
Diskless           X            X          O             X        X
SVID R2V1          X            O          X             X        X
     R2V2          PART         O          X             PART     X
     R2V3          PART         O          X             COMING   X
TCP/IP             X            PART       X             X        X
RPC                X            O          X             X        X
Shared Libraries   O            X          X             X        ?
MMAP(mem mapping)  O            X          O             X        ?
Remote Sys Mgmt    PART         X          O             PART     ?
Net Mgmt           3P           X          O             X        3P
DECnet             X            X          O             3P       3P
VMS Emul           O            X          O             3P       ?
Dist Print Queues  X            X          O             X        X
OLTP               O            X          O             O        ?
Realtime           O            X          O             O        X
Job Control        X            X          O             X        X
Batch              O            X          O             O        O







                                                                Gail Grant
                                                                Eric Jaeger
                                                                11/4/88

    
    RE .4 (??)
    
    The point about LAVC superiority is moot - because NFS will
    do the same things (without ACLs) that you site as VMS strengths.
    I have customers who think NFS performance is better than LAVC.
    To prove that point, compare how many diskless clients each of
    the two environments can support, given similar load constraints.
    
    							  Kris..

RE: .6

 I am not familiar with NFS but I would suggest you are only viewing clusters
in the sense of common file system and it is considerably more than that...

For example:

cluster wide batch
CWPS (5.2)
Volume shadowing for satellites via MIVC & HSCs
failover for disks & servers
etc...

I maybe wrong here but I understand UN*X doesnt support a shared / co-ordinated
file system and that each satellite requires its own directory structure
i.e. no concept of sys$common, this would therefore result in significant
disk storage requirements for the boot (?) node(s).


One other point I forgot to mention in .4...

VMSes view of the world is based on an asynchronous model. Multi-threaded
event driven application can be implemented with operating system support such
as ASTs. UNIX however has a very simplistic synchronous view of the world.
Ironically, the X Window System model is also asynchronous and does not easily
lend itself to the UNIX model. This has certain implications to application
programmers, particularly if they are converting from VWS.

    re .5
    
    	what/where is the SCS note?
    
    	Thanks for the quick response, and the referral from the UNIX
    	notefile (note 353).                                             
                                                            
    	Which notes file is the most appropriate for this discussion,
    	given that DECwindows is often but not always the issue?  I
    	too would love to hear the internals people go head to head
   	on this, but, i would also like to hear from marketing and from
    	sales people.  In my opinion, the issue of functional differences
    	between	Ultrix and VMS needs to be addressed FROM THE
    	CUSTOMERS' POINT OF VIEW.  let's keep this dialogue going!
    
    Thanks, -Anne

I agree wholeheartedly that this is a topic which needs to be addressed from 
a **********TECHNICAL********** standpoint.  Marketing has lots of hype on 
the matter, but the real live discussions from PEOPLE WHO KNOW(i.e. wrote 
it) is sadly missing.  The question has been asked in several notesfiles, 
but everyone keeps saying "pooh, pooh, don't want to discuss, see old notes"

MY CONTENTION IS THAT THE OLD NOTES ARE OUT OF DATE FOR *BOTH* VMS AND ULTRIX.

I'd vote for the discussion to be here, since alot of the stuff affects 
DECwindows, if not directly related(things like LMF are critical in a DW 
environment, as is server failover, availability, clusters vs NFS).  Also, 
there are more engineers in here than most other places.

I think the marketing piece belongs to marketing, but needs to be based on 
the technical comparison, which never gets done for fear of offending VMS or 
ULTRIX.  Some good, but debatable points have been mentioned here.  I'd like 
to see them discussed in full.  I would suggest opening a topic to address 
this in either asimov::marketing or ssgbpm::osf or 
ssgbpm::decwindows_marketing.

I'm going to open a new topic here, if it has not already been done.  Guys in 
Spitbrook(and Palo Alto and Bellevue and CRL), please take alittle time to 
help us out.

g.

    re .8: "what/where is the SCS note?"

    See note 60.* in this conference.

    
    Re: .7
    
    You're right - regarding uniqueness of LAVCs and its monolithic
    file structure for supporting diskless clients.  Yes, you save
    disk space; but what does that buy you? Performance..if all clients
    are swapping DECwindows applications simultaneously, then you're
    in real trouble!  
    
    NFS/UNIX supports a true distributed model by providing all clients
    with separate swap partitions and root file systems.  The penalities
    are the disk space expenditures that you complain about - but the
    benefits are in increased client performance.
    
    Plus NFS is heterogeneous - works on most operating systems including
    VMS (Sun has announced a VMS version).
    LAVCs are very nice for VMS-based systems - with all the rich systems
    management tools and other toys that come with it.
    
    All of the advantges enjoyed by VMS will become moot with the
    retreading of UNIX with staple OSFIX core code ;-) 
    
    BTW:  Did not quite understand your issue with VMS "asynchronousness"
          being better to support the X Window System.
          Could you elaborate?

    George,
    
    care to expalin what the "nonsense" is?
    
    							kris..

re: 220.4
    
    CMKRNL::WALL mentions the comparable characteristics between VMS
    and Ultrix ....  
    
    One of them being PERFORMANCE.
    
    There is no comparison now between our Fastest System - the PMAX
    family - and our most expensive price performance systems - the
    VAX family.
    
    :-)
    
    have a nice day......

    RE: .11
    
    kris,
    
    Not even VMS has come up with the 'cluster common swap file' yet.
    LAVc satellites do not even have have to have their swap file on
    the same SYSTEM as their system disk.  (I know, unsupported,
    unsupported...)
    
    Diskless ULTRIX will still use common copies of the client images.
    Does ULTRIX swap out of the image or copy of image in swap partition?
    I believe it is the latter, or what's a sticky bit for?  If so a
    performance win, diskspace hit for ULTRIX.  But VMS has tools to
    help, permitting multiple system disks on the host(s) of a cluster.
    (performance gain, diskspace hit for VMS)
    
    Interesting point to speculate about is ALL satellites on VMS may
    be beating up on a copy of the a particular RTL if they are all
    running similar but not necessarily identical applications.  But
    maybe I don't remember VMS page fault handling correctly.
    
    -fjp
    

    
    fjp,
    
    I am aware that a lot of neat things can be done on LAVCs with
    crude hackery.  You said it yourself - **unsupported** = some
    customers have a problem with that.
    
    Other problems come with ethernet bandwith constraints.  This
    affects both UNIX and VMS systems.  
    Hopefully the next generation of networking products (eg., FDDI)
    should help.
    
    I have a customer who has tested NFS versus LAVC performance, 
    simulating similar loads.  Their report makes NFS look good.
    Also, there is an internal NFS performance document (written
    by UEG) that sheds some light.  I wish the DECwindows performance
    group would come out with a comprehensive report of their own.
    
    						     Kris...

    I was just pointing out that LAVc's handling of swap space seems
    to map pretty well with ULTRIX's approach.  No crude hackery required.
    
    Putting a satellites swap file on a host disk other than the satellites
    system disk is well documented and supported.  (Even AUTOGEN can
    handle it correctly).
    
    Local page files are handle by both O/S as well.
    
    Putting a satellites swap files on a system other than the host
    or satellite is more slight of hand than crude hackery.  The only
    caveat is that boot order becomes important.
    
    Do you have a pointer to the UEG report on NFS?
    
    Frank

                                                                      
    fjp,                 
    
    My comment about "crude hackery" refers to the implementation
    of the cluster booting mechanism aswell.  I could be all wet
    with that statement!
    
    I have been told by a VMS expert (customer who used to work
    at DIGITAL; who receives VMS sources) that our implementation
    of Cluster support/booting of diskless clients is "very primitive".
    
    I cannot disclose customer's name in a notesfile.
    
    I would like to be educated or proved wrong by our own
    VMS experts.       
    
    re -1
    
    I don't have immediate access to the UEG document on NFS, but
    I remember seeing a pointer in the ULTRIX notesfile.
    
    								Kris..  
    
     

    
    Just dug up some diskless performance data that shows DECwindows
    server/client simulations under ULTRIX 3.0 done by the
    Worksystem Development and Evaluation Group [Ultrix Performance
    Group].
    
    Here is some info that I picked arbitrarily from the charts.
    
    - Calculating Client Task Throughput with an optimized diskless
      server task 0 - 12000 on y axis and 0 - 50 for the number of
      clients on x axis, the following can be observed:
    
      A VAX 8550 server can support 50 clients to do about 8100 tasks
      each.
      A standalone client will do 11000 tasks.
    
      A CMOS microvax will support 24 clients to do 3000 tasks each.
    
    - Calculating Average Client Response Time with an optimized
      VAX 8550, 8 - 20 seconds on y axis and 0 - 50 clients on
      x axis:
    
      The 8550 server will support 50 clients to perform at 13.5sec
      average client response time.  A standalone client will do
      11.5secs and a CMOS MicroVAX will support 24 clients at 16.5sec.
      each.
    
      There is other data relating to Maximum Network Utilization (at
      3 second intervals) and CPU utilization.
    
      This kind of data is needed to compare LAVCs with NFS diskless
      client support in the DECwindows environment.
    
    *BTW:  Also, I have data that shows ULTRIX DECTerm dialog boxes
           and box-downs performing better than VMS. 
           Can this be disputed?
      								Kris..

I don't know what you mean by "crude hackery" involved in the cluster booting
mechanism.  The MOP protocol that the MicroVAX hardware supports is used to
download VMB, which then speaks SCS to the cluster node serving the system
disk.  Once the MOP step is finished, the boot situation is little different
from a VAX with CI-based disks.  I don't see anything particularly hacky or
"very primitive" in that.  This is essentially the same process that Ultrix
uses when it boot diskless--primary bootstrap is loaded by MOP, NFS links
established to the disk server, and boot proceeds normally from there.

--PSW

    
    RE .-1
    
    I am glad that someone thinks "LAVCs are Great" ;-)
    
    I know MOP is used by both NFS and LAVCs for boot and crash management.
    (ULTRIX is about to change that and standardize on an "open" protocol).
    I am *no* VMS expert so I would leave the speculation on "hackery"
    alone.  I have even heard internal VMS users make the same comments.
    Maybe they're all wet like me ;-)
    
    ULTRIX uses NFS to swap and share files from an NFS home directory.
    Does that make a difference with the way LAVC handles it own?
    
    Will LAVCs support multiple environments (layered products, licenced
    products, versions, architectures)?  ULTRIX/NFS can handle all that.
    
    I would still like to see a true study of the differences between
    the DECwindows VMS and ULTRIX in a local area environment.
    
    							 Kris..

�        <<< Note 220.21 by MTA::GRAHAM "take no prisoners, kill them!" >>>
�                           -< some more questions... >-
�     
�     ULTRIX uses NFS to swap and share files from an NFS home directory.
�     Does that make a difference with the way LAVC handles it own?

    With VMS V5.0 and the lifting of many restrictions, you can pretty much
    think of all types of VAXclusters being equal (at this level of
    discussion). VAXclusters allow you to serve disks located anywhere
    on the VAXcluster to the entire cluster, and you can then mount them
    on an arbitrary set of cluster nodes. So pretty much, anything can be
    located anywhere - the most notable and currently limiting restricition
    is that the dump file must be located on the system disk. I expect this
    restriction to be relaxed in the near future. And of course, if you want
    to deal with disks not in your cluster we have the DFS product, which
    can be likened to "NFS for VMS", but with more robust file locking.

�     Will LAVCs support multiple environments (layered products, licenced
�     products, versions, architectures)?  ULTRIX/NFS can handle all that.

    They support them as well as VMS does :-). In my opinion, this is
    current weak spot in VMS - supporting multiple installed versions of
    products. Those of us who don't mind fidling with a few logical names
    rarely have a problem with this, but it is not officially supported. If
    base VMS supported this, so would VAXclusters. Are there specific
    features in NFS which extend the UNIX ability to do this, or is this
    just inherently supported by UNIX? 
     
    I'm not a UNIX expert, so I hope I haven't mis-stated the case for UNIX.
    I do consider myself very knowlegeable about VAXclusters, which I veiw
    as very good stuff...Ken 

    VMS will permit multiple system disks for different environments
    for a group of satellites.  The catch is it has to be a disk.  ULTRIX
    wins in that you can build your environment on a partition and save
    a bit of space, support multiple environments with a smaller host
    configuration.                                             
    
    I'm not sure how easy it is for ULTRIX managers to support the random
    3rd P. layered products on a diskless systems environment.  Few
    vendors seem to support Digital's added value SW installation procedure
    (setld).  VMSINSTAL gets pretty broad use.
    
    I didn't think DFS supported any shared simultaneous access to files,
    robust or otherwise.  But I haven't seen info on the latest release.
                                                               
    fjp

    
    RE .22
    
    NFS is "stateless", ie, it allows system and user files to be put
    anywhere on the LAN, and users see the entire LAN as one system.
    vis..."the network is the computer" ;-)
    
    For instance one of our local groups here ported STSC APL to PMAX/DEC-
    windows, and the availability of NFS mounts has made it possible
    to inter-operate and test different versions of software on different
    workstations running different versions of software (including VMS).
    
    								Kris..

RE: .24

DFS for VMS is a closer equivalent to NFS for Unix than is a LAVC.  Members of
a LAVC share much more (security domain, lock management domain, queues) than
just files, and therefore a LAVC is at the same time a more tightly integrated
and more restrictive domain than a set of Unix nodes joined by NFS.  DFS, like
NFS, is "stateless," as you put it.  The individual nodes have more autonomy
than nodes in a LAVC, which can be both an advantage and a disadvantage,
depending on what you are trying to accomplish.

--PSW

>  Plus NFS is heterogeneous - works on most operating systems including
>    VMS (Sun has announced a VMS version).

   Gee, it's good to see that SUN is catching up. VMS has had an NFS server
   in the UCX product for some time now....

   James 
  

    
>    I know MOP is used by both NFS and LAVCs for boot and crash management.
>    (ULTRIX is about to change that and standardize on an "open" protocol).
 

    If `Ultrix is about to change that', who's gonna pay to upgrade all the
    boot roms, and ethenet controller roms in all the Ultrix systems to use
    this new 'open' protocol. MOP is implemented on the controller, changing
    it is a non-trivial task.

    James

    
    >> Gee, it's good to see that SUN is catching up. VMS has had an
       NFS server 
       VMS has had an NFS server an the UCX product for some time now....
                                 
    
    But the VMS-NFS server doesn't serve VMS machines!!  Why?!!  The same
    old politics?!
    
    Sun's version, I believe will do just that.  Maybe (*just maybe*),
    at that time, DIGITAL will rethink its "lock you in" politics.
    
    								Kris..
    

Right; obviously it isn't done that way.  What might be true (though at
this state nothing has been decided and no details have been worked out)
is that *future* products could offer a choice of boot protocols, which
could include MOP, standard protocols (IEEE 802.1E) and some appropriate
selection of the Arpa protocols (bootp, tftp, etc.).

	paul

    
    UEG actually has a new "open" protocol agreed upon. Could be in   
    an ULTRIX product faster than you think.
    
    Can't disclose any details in here without the product manager's
    permission.                                                     
    				
    								Kris..

RE .24

I think that Ken made it appear that various logical names need to be set up to
use multiple versions of a product on VMS in a LAVC. While this is the normal
way of handling multiple versions, it is also possible to place the relevant
files in a system-specific directory to get the same effect.

The reason why this would not often be done is because it makes the cluster less
homogenous. The implication being that only the intended users could use that
version of the product on that system. The method Ken alluded to does not cause
suprises to users logging onto the 'wrong' machine.

re: 27

Actually mop is only used to down load the load assist agent, which starts
up a Virtual circuit on a datagram service (Ain't that cute?). This is 
essentially SCS talking. Then the load and crash paths are taken through SCS
on a LAVC, not via MOP. ULTRIX uses MOP for download/upload operations
fully at this time.

Jeff