-
Type:
Sub-task
-
Status: Closed (View Workflow)
-
Priority:
Minor
-
Resolution: Fixed
-
Affects Version/s: 0.20 Larks
-
Fix Version/s: 0.20 Larks
-
Labels:None
By default C3P0 has PreparedStatement cache size set to 0 which makes PreparedStatement objects to be created each time and to be actually prepared each time.
We want to configure C3P0 to have that pool. As a result of this task we need a report of how the performance was changing depending on the size of the statement cache.
This task should be started only after performance tests are written and automated.
Stanislav Bashkyrtsev can also leave the query cache?
I looked at the time of the request to the main page with cache (1.19-1.50 seconds) and without query cache (2.5-3 seconds)
PR preparedStatement
CQ cashe of queries
count of requests 4800
| config | min response duration(ms) | max response duration(ms) | average duration(ms) |
|---|---|---|---|
| with PR without CQ | 83 | 17964 | 1035 |
| without PR with CQ | 83 | 11807 | 799 |
| without PR without CQ | 82 | 16384 | 1128 |
| with PR with CQ | 83 | 12073 | 955 |
RESULT:
is configured: without PR with CQ
query_cache_limit=1M
thread_cache_size=50
query_cache_size=100M
query_cache_type=1
Main thing to test here is to verify if we have gained some performanc benefits out of it
Can't be tested; we have no previous results (before the fix)
ok, removed previosly changes in DB configuration. Add parameter to connection string. For check result: SHOW GLOBAL STATUS LIKE '%prepared%';