Re: Oracle Performance on HP vs Sun ?

From: Todd Thiemann <tthieman_at_cup.hp.com>
Date: Tue, 15 Mar 1994 18:55:17 GMT
Message-ID: <CMpz85.MoF_at_cup.hp.com>


Tim Read - Sun Linlithgow Snr SE and DB Amb. (timr_at_cerrera.uk.sun.com) wrote: Tim-

A bit of clarification that might help here. I'm trying to help answer the original question, which was developing a comparison between the HP9000 E35 and the SS10:51. The original request (if I recall correctly) was for TPS-type metrics, so that is what I tried to help estimate. This is similar to the "Estimated Transactions/sec" that Sun provides on the SPARCserver FAMILY data sheet.

TPC-A and TPC-C are apples and oranges, but you should be able to get a rough estimation of a system's OLTP performance by the ratio I mentioned. That ratio is tps, not tpsA or tpmC, but the rough estimate of transactions per second, similar to what Sun publishes on the data sheet (something I haven't seen on HP literature). I'm fully cognizant of the rules concerning the mention of TPC numbers - that is why I always included cost per tps or tpm when I quoted those numbers. What I mentioned is an extrapolation that compares develops a ratio of apples to oranges that could then be consistently applied in this particular situation because the TPC-Cs both used Informix and the TPC-A for the SS1000 used Oracle. The result is something that might be compared against the Sun tps estimate on the Sun data sheet.

The T500 was introduced in November 1993 and has been shipping since that time. I think the reason that you mention a July availability is because the TPC full disclosure report mentioned that date as the availability for THAT configuration.

I think the reference you make to "kick anyones butt" came from another netter, not me. I appreciate your responding to that, but I wanted to help the original poster solve the problem that initiated this discussion.

I hope this matter is laid to rest and regards,

Todd Received on Tue Mar 15 1994 - 19:55:17 CET

Original text of this message