Re: Article about supposed "murky" future for Oracle

From: Douglas Hawthorne <douglashawthorne_at_yahoo.com.au>
Date: Thu, 01 Apr 2004 23:14:22 GMT
Message-ID: <in1bc.135794$Wa.93961@news-server.bigpond.net.au>

> Versioning has its own problem too. I am sure you would have gone thru
this:-
>
>

http://www-3.ibm.com/software/data/pubs/papers/readconsistency/readconsistency.pdf

On p.8 of that report,

"Benchmarks Prove it
"TPC-H
"In data warehousing benchmarks, namely TPC-H, readers and writers do not

run concurrently. That is, the benchmark is designed to run query streams and update streams at different times so multi-version read consistency does not play a role."

At http://www.tpc.org/tpch/results/tpch_perf_results.asp , there are the following results:

The 100GB results has DB2 8.1 in the top three (3), followed by six (6) SQL Server with one (1) Sybase in the middle. The 300GB results has DB2 8.1 and 7.2 in the top three (3), followed by three (3) SQL Server then two (2) Sybase. The 1,000GB results has Oracle 10G and 9i in the top three (3), followed by two (2) DB2 (7.2 then 8.1), another Oracle, then two (2) SQL Server. The 3,000GB results has Oracle 10G and 9i in the top six (6), followed by DB2 7.2, then Oracle 9i.
The 10,000GB results has DB2 8.1 followed by Oracle 10G.

It should be noted that these results include clustered and non-clustered set-ups. The 10,000GB had DB2 on clustered with the Oracle on non-clustered.

Douglas Hawthorne Received on Thu Apr 01 2004 - 17:14:22 CST