Re: DB Buffer Cache Size

don_at_burleson.cc (Don Burleson) wrote:
> > > Absolutely. A cache has to be managed. Make it too big, and
> > > management becomes all you do, and never mind serving up data.
>
> As some would say, LET'S PROVE IT!
>
> For example, why do ALL the 10g "world record" benchmarks use over 100
> gig data caches?

Hey, the world-record space shuttle runs off liquid hydrogen, liquid oxygen, and solid rockets. Maybe that means Greyhound should convert their bus fleet to do the same.

> I worked with one of these TPCC benchmarks and ran
> repeatable timings. Up to the point where the working set was cached,
> the benefit of a larger data cache outweighed the LIO overhead.

How do you identify exactly what the "working set" is and know when it is cached?

>
> But don't take my word for it, see for yourself:
>
> http://www.tpc.org/results/FDR/TPCC/HP%20Integrity%20rx5670%20Cluster%206
> 4P_FDR.pdf
>
> db_cache_size = 4000M
> db_recycle_cache_size = 500M
> db_8k_cache_size = 200M
> db_16k_cache_size = 4056M
> db_2k_cache_size = 35430M

Maybe I forgot how to do math, but that doesn't seem to add up to over 100G.

Xho

-- 
-------------------- http://NewsReader.Com/ --------------------
Usenet Newsgroup Service                        $9.95/Month 30GB