Re: How to get acceptable/repeatable performance on huge table

[Peter Stridsberg]

show processlogs for 15 minutes reveals "Copying to tmp table".

Data sizes are huge, so no wonder:
mysql> SHOW TABLE STATUS LIKE 'UserVisits'\G
*************************** 1. row ***************************
Name: UserVisits
Engine: InnoDB
Version: 10
Row_format: Compact
Rows: 229781925
Avg_row_length: 137
Data_length: 31576817664
Max_data_length: 0
Index_length: 0
Data_free: 7340032
Auto_increment: NULL
Create_time: 2012-01-24 18:58:03
Update_time: NULL
Check_time: NULL
Collation: utf8_general_ci
Checksum: NULL
Create_options:
Comment:

When possible, I will add another index (adId, Uid) hoping that this will avoid temporary table and that the query could access the index alone with no need to access actual data (am I really wrong that sizes of other data columns would then not matter to this query?)