Hi.
Well, your are right, indexes are not used at all, so ignore
everything about key_buffer and indexes I said, regarding this issue
(but key_buffer=16MG seems small with 1GB anyhow ;-)
Depending on how often you need the query, and how much columns are
involved, you may want to consider to
I have one query, that basically reads from one table of 13 million rows,
groups it by two fields, and inserts into a rollup table. The result set
ends up being about 31,000 rows. This runs in about 2 minutes which is
acceptable. However, I add an additional group by field, which will probably
Hello.
First, key_buffer=16M seems a bit low for me with 13 million rows
(well, depends on the indexes, and how much free memory you have, of
course).
If I am not mistaken, sort_buffer/tmp_table_size are the one mostly
needed for this query, I think. Btw, you can get more info about the
query
Benjamin Pflugmann [EMAIL PROTECTED] writes:
Hello.
First, key_buffer=16M seems a bit low for me with 13 million rows
(well, depends on the indexes, and how much free memory you have, of
course).
1 gig, plenty free.
If I am not mistaken, sort_buffer/tmp_table_size are the one mostly