Re: Oracle and OpenVMS (was: Rename Column in 9iR2)

On 2002.05.23 22:43 "Brown, Pedr" wrote:
> OpenVMS is a rock solid plafform, we run 120+ 817 databases (spread
> across
> 20 nodes) with very few platform issues. The downside is some
> functionality
> doesn't work too well (like MTS) and on the benchmarks we've done
> inhouse
> Oracle/Tru64 has always been roughly 2x as fast as Oracle/OpenVMS on
> the
> same hardware configuration.

That is because VMS has an old phylosophy of using CPU modes for separation
of privileged parts of the program and, partly, because of the logical names.
Logical names are implemented as devices. Turn on IO tracing with the "set watch"
command and you'll see that every access to a logical name table (process, batch, group,
system) causes an IO to happen. VMS is an old system with a huuuuuuuge kernel
which is better suited for TP-monitor type of applications then to an intense
myriad of small processes so characteristic for Unix. The only way to beat unix is
not to use two-task architecture at all, because IO is extremely expensive on VMS
but to use so called single-task architecture (S:) which used to be available on VMS
long time ago. Also, turn off any OS caching (Files11 started doing that as of VMS 6)
as Oracle does it's own caching and VMS caching only interferes and wastes the necessary
memory. Also, make sure that swapper is not too active. You need to tune the memory
variables (FREELIM, BORROWLIMIT,GROWLIMIT) extremely carefully because an overactive
swapper can really kill a VMS machine. Also, tune the MPW (modified page writer)
and make sure that the non-paged pool is sufficiently large. I was able to beat
an HP 735 with HP-UX 7 with a MicroVAX 3900 with VMS 5.0 (a long, long time ago in a
galaxy far, far away when there was SQL*Net V1.0 which was started as a process called
"orasrv"). Im sure that modern Alpha machines can beat the crap out of OSF/1 microkernel
Unix. Consult an old lore by Clay Prestia and Bruce Ellis (where is Billy Bitsandbites
when youneed him?). I used to teach people how to tune VMS (I stopped using it when
the world was much younger and the version was OpenVMS 5.5-2) and I know that VMS
can be very, very highly tuned. If tuned properly, I'm sure it can beat any Unix on
a comparably fast HW.

>
> -----Original Message-----
> Sent: Friday, 24 May 2002 1:33
> To: Multiple recipients of list ORACLE-L
>
>
> As a side note, please please please become familiar with OpenVMS and
> more
> importantly Oracle on OpenVMS before tackling this. At least as far

Left OpenVMS for Irix 5.3 and never looked back. After all, I had to resign
because I advised my boss to buy MIPS R3300 based DECSystems 5800 with Ultrix.
In slightly less then a year, a DEC salesman stopped by and told my boss something
like:" No upgrade, no trade in, no transition, no support for Ultrix. Throw your
boxes away and buy the new and shiny alphas". Aftere that, I was asked to resign.
I am actually glad that DEC has perished. They certainly deserved it! No more
OpenVMS for me. I'll stick with the mainstream. Unix rulez! OpenVMS was a nice system, DEC products were beautifully engineered but their
marketing sucked big time. They fully deserved what happened to them.

-- 
Mladen Gogala
-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: Mladen Gogala
  INET: mgogala_at_adelphia.net

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).