Migrating to Unix: SGI vs. SUN

mathog at seqaxp.bio.caltech.edu mathog at seqaxp.bio.caltech.edu
Fri Sep 13 13:47:35 EST 1996

In article <1996Sep12.155933 at bphvax>, caal at bphvax.biophysics.rochester.edu (Charles A. Alexander) writes:
>We're presently evaluating the pros and cons of migrating our GCG 
>package to a UNIX platform.  It seems like a lot of the freeware 
>biocomputing apps are being written for UNIX and though I love my VAX 
>and VMS System Management, it seems like I would be doing my users a 
>disservice by not providing at least the option of other software.
>We also want to provide molecular modelling tools for the people we 
>support and it seems that SGI and SUNS have a lot of tools available for 
>these applications as well.

If you're going to be doing molecular modeling you need an SGI that isn't
going to be used for anything else.  It pretty much has to be an SGI 
because all the molecular modeling software runs on them, but only a few
packages run on each of the other platforms.  Until such time as all of
this software gets rewritten to OpenGL I don't see much likelihood of that

I don't know about your users, but 95% of mine hate Unix as much as VMS.
They aren't so fond of GCG either ("You mean we need to remember the name
of the program to run?  It uses a what, command line interface???") but
when they use it enough, generally come to appreciate that it works quite 
well.  Still, for the last couple of years, up until about a year ago, they
were demanding something that was Mac or Windows native (as appropriate). 
Now, they pretty much expect everything to pop up in Netscape.  Lord only
knows what will be trendy next year :-(.

Bottom line, most of them won't like GCG on Unix any more than they like
it now on VMS. after all it's the same thing except for the minor
interface change of "/" vs. "-" and the case sensitivity problems. 

You will find that more software runs out of the box on Unix systems,
primarily because somebody else has already made the OS specific changes
that the programs need.  Well written ANSI C compliant code runs on
anything, but you won't find as much of that in the public domain as you
might wish.  I've ported scads of "Unix" programs to OpenVMS, rarely is it
much work, there are usually fewer "ifdefs" for OpenVMS than between the
Unix variants.  For instance, I added practically nothing to get the last
version of the NCBI toolbox running.  In other words, for most things, the
fact that a program was written on a Unix system is irrelevant - it just
processes data, and it will do that with small changes on any other
platform.  However, I do still run into programs that are not in ANSI C, or
worse, are Sun or SGI specific (usually for no good reason, just lazy
programming.) On the other hand, WNT is clearly winning the OS wars, and if
you have a lot of PCs around (we don't, almost all Macs) you *will*
eventually be running WNT servers. 

If you have enough cash, go and buy that SGI for modeling.  Then buy an
Alpha, go with OpenVMS (if you don't want to retrain your users) or Digital
Unix (if having a bit more software matters).  Then later on you can shift
the thing over to WNT if you need to.  That should keep you in business for
about 5 years, which is the prediction limit anyway.  If you want a pure
SGI shop (which would simplify your life) buy a second one to run GCG on,
but get a server with minimal graphics (you don't want people modeling and
database crunching on the same machine).   Hint, since SGI changed
around their licenses you should more seriously consider getting an SMP
capable machine, since otherwise you have to buy extra licenses for extra
capacity.  SMP works great on our two processor 2100.

Odd that in industry HP seems to be the most common Unix platform, but they
are rare in academia.  GCG doesn't even support it, either not enough
demand, or maybe some major incompatibility with the GCG software? 


David Mathog
mathog at seqaxp.bio.caltech.edu
Manager, sequence analysis facility, biology division, Caltech 

More information about the Info-gcg mailing list

Send comments to us at biosci-help [At] net.bio.net