This isn’t actually new — it has been out for a few releases. The mk-query-digest tool from Maatkit now outputs information about each class of queries’ variance-to-mean ratio. The new output goes in a couple of places, including perhaps most usefully the “profile” report. Here’s an example from a real MySQL system:

# Profile
# Rank Query ID           Response time    Calls R/Call Apdx V/M   Item
# ==== ================== ================ ===== ====== ==== ===== =======
#    1 0xBFCF8E3F293F6466 11256.3618 68.1% 78069 0.1442 1.00  0.21 SELECT [redacted]
#    2 0x620B8CAB2B1C76EC  2029.4730 12.3% 14415 0.1408 1.00  0.21 SELECT [redacted]
#    3 0xB90978440CC11CC7  1345.3445  8.1%  3520 0.3822 1.00  0.00 SHOW STATUS
#    4 0xCB73D6B5B031B4CF  1341.6432  8.1%  3509 0.3823 1.00  0.00 SHOW STATUS
# MISC 0xMISC               560.7556  3.4% 23930 0.0234   NS   0.0 <17 …

The best overall method of performance optimization is optimization for response time. Users care about response time, not load average or cache hit ratios. The job of a system is to accept some request and do the required work, and deliver a result. The time elapsed between the request and the result is the response time.

Methods of Response Time Optimization

Not all optimization methods are created equal. Here are a few I see commonly.

• No method. Most people simply have no method of performance optimization at all. They just look for things that look “bad” and try to make them look “better.” In the MySQL database world, the classic example is trying to improve a cache hit ratio. This is utter folly, and doesn’t become any less stupid no matter how many times it is taught and repeated.
• Server Load Reduction. One step up from that is to try …

Last week I attended the Enterprise LAMP Summit and Camp in Nashville, Tennessee. I enjoyed the event and met or reconnected with a lot of great people. I was glad to be able to spend time with some folks from the Postgres community. My own sessions focused on MySQL.

During the Summit I tried to help people understand how to think about performance, and made the case that the Percona versions of the MySQL server are not only the highest-performance available, but uniquely provide the instrumentation necessary to follow a disciplined performance optimization process such as Method R or Goal-Driven Performance Optimization.

At the Camp the next day, there were several sessions on MySQL. My talk was later in the day, so I elected to skip …