Re: Why INSERT performance has not increase in sub-table?
Posted by:
Dexter Yao
Date: November 03, 2014 11:56AM
Thank you for your reply, Rick.
I change the test method in TestB, insert the records in order, e.g, insert 100000 records into testb_1, then testb_2, then tesetb_3, etc. In such way, the insertion performance is close to one table test, so i guess it is ‘Cache’ problem.
You told me there at least 3 vars, innodb_buffer_pool_size, innodb_flush_log_at_trx_commit, and ?
When i tried the throughput, i find when the connection number is about 50, the throughput reach the max. Is there something wrong with this number or my method?
Just now i change the innodb_buffer_pool_size from 128M(default value) to 1G, the performance improved a lot. And i found on the problem “one vs many”, it seems has no reason to choose the latter, or some other thing i did not consider.
PS: Sorry for the poor english, wish u can understand what i mean. :)