Re: SQL Cluster Performane worse than MyISAM.

[Andrew Mcleod]

Thank you for your reply Mikael.

It is good to know that future releases are looking at dealing with these types of queries better, I'll be hanging out for the version 5 release.

Below is an explain on the query:

SELECT b.*, m.name FROM iBB_blog_blogs b LEFT JOIN iBB_members m ON (b.member_id=m.id) WHERE b.blog_private=0 AND blog_name IS NOT NULL ORDER BY b.blog_pinned DESC, blog_last_date desc LIMIT 0,20
table type possible_keys key key_len ref rows Extra
b ref blog_private blog_private 1 const 10 Using where; Using filesort
m eq_ref PRIMARY PRIMARY 3 boostcruising.b.member_id 1
MySQL time: 5.12795

I've been using replication with a manual switch over method if the primary host goes down for a significant period of time - and in regards to performance, using the HEAP engine is not an option as I do not want to loose data when the daemon restarts - using the NDB engine seemed like a perfect solution...

-----