Re: Good News for MS Windows users: Your favorite database is here..

On Wed, 23 May 2001, Nuno Souto wrote:
>On 22 May 2001 12:45:30 -0400, Haider Rizvi <haider_at_no_uce.ca.ibm.com>
>wrote:
>
>>nsouto_at_nsw.bigpond.net.au.nospam (Nuno Souto) writes:
>>
>><bunch of stuff deleted>
>>
>>> When was the last time you tried to saturate 12 CPU's on a Sequent
>>> with a single DB2 parallel query?
>>
>>Look at DB2's 300 GB TPC-H results from a year ago,
>>http://www.tpc.org/tpch/results/h-ttperf.idc. DB2 published three
>>results on numa-q (previously known as Sequent). Results are on 64, 48
>>and 32 cpus, and they scale well just look at the numbers! And I can
>>promise that even for the 64 cpu runs, all the cpus were saturated,
>>the scalability kind of proves that. Q.E.D.
>
>Not QED at all. I said a single parallel query, that's not exactly
>the same as a TPC-H? Or can't you spot the difference?

I'm well impressed. More posts on a single thread than I make in all the newsgroups I subscribe to, and in every one he's indisputably right. Of course, the phrase "splitting hairs" came to mind, unbidden, more than once, along with "life donor".

If I were a life donor interested in splitting hairs, I could, for instance, point out that most of us (even those who work for hardware vendors) don't really have a 12-way Sequent box lying around to saturate with a single parallel query (or a can of DB2 for that matter!) just to prove whatever point it was you were trying to make.

As someone who doesn't work with either Sequent, DB2 or Oracle, a series of TPC-H benchmark records using Sequent and DB2 seems more convincing on scalability and parallelism than some vague and unspecified claim relating to saturating 12 CPUs.

I also couldn't really care about the scalability of a single, arbitrary query which might be life or death for you, because I don't work on your system. If I did, I'd possibly feel differently. Although, frankly, I can't see how any site where you worked would need anybody else around except to make you coffee or bask in your radiance, and I'm crap at both.

Things like this that remind me why cross-posters should be hung, drawn and quartered (ipse dixit). Received on Sat Jul 21 2001 - 18:05:52 CDT