I’m trying to setup tpcc-mysql to benchmark a server. Running ./tpcc_load but it’s very slow…ran in parallel, still slow. Is there anything I’m not doing right.
Commands are being run from a VM with 5G mem and 4 cpus, nothing else is running on that VM. The server in question is a HP G8 server, also idle. Any thoughts?
Sorry to waste your attention. I skipped to mention that two servers were on different networks, and bandwidth was limited. I moved the tpcc_load onto same network as G8, and load rate increased significantly. I’m populating 1000W, if anyone wants sample data, let me know.
It’s hard to tell why there is behavior like this.
Possible causes are: cold/hot cache, size of transactions, innodb checkpoints, other server activity…
The picture clearly shows us that peak values are regular.
To be able to answer exactly what’s going on here - I would run pt-stalk ([url]http://www.percona.com/doc/percona-toolkit/2.2/pt-stalk.html[/url]) while this test is running and then analyze its data collected at the peak and drop moments.