Keeping up with HW and SW technology? Possible?

Keeping up with HW and SW technology? Possible?

Post by John A. Johnst » Sat, 02 Jan 1993 02:09:25



A dilemma we're constantly facing, and looking for the *real* answer to it,
or at least pointers to a potential resolution.

Basic information:
 - We provide a database using Sybase (the server and front end interface)
 - The database development and maintenance is done on Sun hardware

The three ingredients involved in this environment; Solaris (SunOS),
Sun Hardware,  Sybase Are horribly intertwined.

Eg.  To upgrade a dozen workstations coming of age this year, the Sun
LX fills our hardware performance needs.  The LX *requires* an
immediate migration to Solaris 2.1, Sybase does not work with Solaris
2.1, and won't for a few months at least, then we need to verify our
apps are ported ok to the revised Sybase, and Solaris, which in turn
takes attention away from the next generation of our product, and time.

This leaves us the "option" of not following our best interest hardware
wise, possibly to spend much more than we need with the Sun10, or buy
yestertech and go with the SunSparc2, or reamin status quo.

If we remain status-quo, and hold off on our hardware migration, we
basically would lose up to 1/2 year of the actual useful life of the LX
workstation (where buying a workstation near the beginning of it's
product life increases by that much, how far down the road it will be
"state of the art") and lose more of any useful re-sale value of the
Sparc 1's, as they grow further behind.

The basic solution seems apparent, just slow down the migration path,
and expect to be 1/2 to 1 full year behind the state-of-the art as the
various major and minor releases of Solaris and Sybase dictate.  This
will cost something in hardware value to us, but we are happy.

The resluting problem is a customer...  They want what we have our
product, but are not at all happy buying the last generation of
hardware, when the latest out is the better price/performer.

There is no clear win-win scenario.  Software schedules from Sun,
Sybase and even ourselves are usually not right on any schedule.

What do *BIG* and *Medium* developers do to plan/prepare/combat this
hw/sw interconnection?  I'd classify us as tiny, but facing the same
problems as the biggies, but on a smaller $ budget..

Any and all input appreiciated, and to be summarized.
-johnj

 
 
 

Keeping up with HW and SW technology? Possible?

Post by Barry Margol » Sat, 02 Jan 1993 17:06:11



Quote:>If we remain status-quo, and hold off on our hardware migration, we
>basically would lose up to 1/2 year of the actual useful life of the LX
>workstation (where buying a workstation near the beginning of it's
>product life increases by that much, how far down the road it will be
>"state of the art") and lose more of any useful re-sale value of the
>Sparc 1's, as they grow further behind.

You're assuming that you'll eventually upgrade to LX's.  However, I suspect
that by the time you're ready to upgrade to Solaris 2.1, there will be an
even newer workstation available (new models seem to come out every 6-12
months these days).  So you'll still be able to upgrade to a product at the
beginning of its life cycle.

In general, trying to optimize stuff like this is a no-win situation.
Consider the contrary problem of someone who never buys anything, because
there's always something better coming "real soon now".  In an industry
that moves as rapidly as ours, you're generally winning if you get hardware
for which maintenance is still available :-)
--
Barry Margolin
System Manager, Thinking Machines Corp.