Details
-
Bug
-
Status: Done
-
Medium
-
Resolution: Fixed
-
8.0.21-12
-
None
Description
I am pretty much aware that select count is a wrong way for performance measurement,
but in this case MyRocks is so slow, so most likely something is wrong here.
I compare select count from sbtest1 for both MyRocks and InnoDB under comparable condition, with 6GB allocated for cache and in total 10GB of memory available for mysqld process.
query is run after cold start.
in InnoDB:
mysql> select count from sbtest1; ---------- count ---------- 50000000 ---------- 1 row in set (36.77 sec)
in MyRocks:
mysql> select count from sbtest1; ---------- count ---------- 50000000 ---------- 1 row in set (1 hour 37 min 29.28 sec)