|
On 14/04/2014 4:30 PM, Ryan Johnson
wrote:
FYI, here's a plot of performance over time. Each point in the graph is throughput (in tps) over a 10-second measurement (~20 minutes total), against a 12 WH TPC-C dataset with 24 clients banging on it. I issued a pg_ctl stop/start pair between each run:  Updated result: SI definitely does not suffer any performance loss over time, but it's not clear what is wrong with SSI: deleting pg_serial/* had exactly zero impact on performance. The two near-zero results for SSI are both cases where pg_serial/ ballooned to 8GB during the run (I saw two others tonight during testing). The one big dip for SI was a hang (the second such hang tonight).
|
