June 20, 2013

Post: To SQL_CALC_FOUND_ROWS or not to SQL_CALC_FOUND_ROWS?

…), d=MD5($i)”); } First of all, let’s try to perform some query on this table using indexed column b in where clause: mysql… used index scan inly which is much faster here. Just to be objective I’ve tried to perform this test without indexes (full scan) and with index on b column. Results were following…

Post: Using CHAR keys for joins, how much is the overhead ?

… simple MyISAM table and join query performed on INT fields: CREATE TABLE `intjoin` ( `…this to shorten some internal buffers MySQL has to allocate for key … character based keys – uuid, sha1/md5 based etc. Latin1 encoding indeed … to be different as longer indexes expected to have much worse cache…

Post: To UUID or not to UUID ?

… separate problem which can be fixed in MySQL 5.1 Data Clustering This again applies…this little case we have about 200 times performance difference which is worth to consider Can… point in time physically local in BTREE index. This is actually the reason why UUID… same post – hashes if it is SHA1, MD5