On Mon, 2003-03-03 at 07:41, cbbrowne@cbbrowne.com wrote:
> > (In fact, they seem to think they are testing kernel performance, not
> > database performance, which strikes me as rather bizarre. But anyway.)
>
> That may be a terminology thing; the main SAP-DB process is called the
> "kernel," and it's more than likely that the "SAP-DB Kernel" is the sense in
> which the term is being used.
Actually, I believe the reason the benchmark was developed was to
provide a workload for optimizing high-end Linux kernel performance
(with the inference being that SAP-DB is pretty close to Oracle, Oracle
performance is important for enterprise deployment of Linux, and
therefore optimizing the kernel's handling of SAP-DB running TPC
benchmarks will tend to improve the kernel's performance running
Oracle/DB2/etc.) So when they mean "kernel", I think they really mean
"kernel".
That's not to say that the benchmark wouldn't be useful for doing other
stuff, like pure database benchmarks (as long as its a valid
implementation of TPC-C (or TPC-H, etc.), it should be fine...)
A research group at the university I attend (www.queensu.ca) expressed
some interested in a TPC-C implementation for PostgreSQL, so I was
planning to port the OSDL TPC-C implementation to PostgreSQL.
Unfortunately, I got sidetracked for a couple reasons: (1) lack of time
(2) increasing awareness of just how boring writing benchmark apps is
:-) (3) distaste for ODBC. While I'd like to get some time to do the
port in the future, that shouldn't stop anyone else from doing so in the
mean time :-)
Cheers,
Neil
--
Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC
