The latest from the MySQL community

It is hard to believe a year has passed since the last MySQL Conference. It also seems like so much has happened since then.

From a community perspective one major difference from last year is how many of the active community members have become MySQL employees in the last year. Congratulations to all the folks who joined the force.

It is good to see all of these familiar faces, and to meet new ones. The speakers lounge was buzzing with activity, and the conference halls will be shortly as things get rolling.

The MySQL Conference has started. It will be a long week, still yet to prepare my own presentation for tomorrow. Old friendships already renewed, plenty of faces to names already, and we have yet to hit the first session.

The MySQL Conference kicks off this week in Santa Clara through Thursday.  We have tutorials and various customer and partner meetings today.  Some of the tutorials are sold out, but you can still register on site.  The main conference program starts Tuesday and there will be news announcements each day.

There's been a couple of late additions to the program which should be pretty cool, including a session called HA MySQL on EC2.  It sounds like a lot of acronyms, but it's very interesting stuff.  It's all about using Amazon's Elastic Compute Cloud to create a virtual hosted scale out solution to high availability MySQL applications that you pay for "by the drink."  So if you imagine a web application which can have huge swings in demand and capacity, you can simply provision more computing power at …

HA is for High Availability. And, "real" depends on what you mean. People used to measure availability in "nines". A typical demanding level, used in telecommunications, is called five-nines because it represents 99.999% availability. Over one year span, this leaves you with about 5 minutes of total allowed downtime. To have a better idea what it means, you have to assume the reliability of your system, i.e. how often it can fail or be brought down--for any reason: software, hardware, or maintenance. Say, it is once per week. This is 52 times during a year. With that, your time to recover is narrowed down to 6 seconds. Very likely, to have some additional safe margin, you would like to recover in 1-2 seconds. The game becomes serious. You know there is no way you can physically repair any computer in 1-2 seconds. You need to have redundant hardware and to have at least a hot-standby configuration (active/standby) to perform failovers.

In the …

SolidDB
Transactional, ACID compliant Engine
Optimistic and Pessimistic Concurrency Control (Innodb uses Pessimistic)
Configurable Durability
Support transaction isolation on a per table basis


PBXT
Transactional not fully ACID compliant
Recovers from a crash faster then myISAM
Designed for Heavy INSERT usage

Memcache Engine
Tables with Key Pairs
Insert into memcache through mysql and use the memcache client to select against it. Incredibly fast on insert, very nasty bug in memcache client C-API that more then 100 connections, lockups occure.
Multiple mysql servers can share the same data
Monitoring memcache servers with table data in mysql

AWS Engine
Store data in S3 - cheap way to store data not really going to use
Alpha

HTTP Engine
Generic web service Engine
SELECT - GET …

Oooh in 5.1 these are now system tables. Why should you care? The slow query log requires a full restart of mysql in previous versions of mysql. Now it's just a flag that can be turned on and queried against from the CSV storage engine.

I'm in the mysql uc tutorial on an indepth look at mysql 5.1. A new tool is out to replace my various benchmark tools called mysqlslap.


mysqlslap --concurrency=5 --iterations=5 --query=query.sql --create=create.sql --delimiter=";"


Very cool stuff to test various engines with a sites specific query type and uses.

update


http://hg.tangent.org/bbench?f=f5abeffaf040;file=load_run.sh


There are many things we can do on daily basis which can make our lives better and make future better for our kids and ourselves. Most of these things are overlooked or not stressed enough. Lets start with cheap way to save some money and make world a better place.

Replace all the bulbs [...]

I got tired of going all over the place to get information about mysqld variables, so I decided to make a summary of most of them (for version 5.0.27 linux). (Due to lack of horizontal space, I had to make two tables — the second one containing a short description of the variable.)

Corrections, additions, and amplifications are welcome. Enjoy!

Table 1: MySQL Variables

Saturday and Sunday I attended SDForu.m's 2nd silicon valley ruby conference.

I attended to scope out what the Ruby community was up to, see what the skillsets are like, etc. I've been trying to get into Ruby myself for a few months. Slowly getting there.

The conference was at The Tech in san jose. Was a nice cozy room. One room, one track, speakers came up one after the other. Most of it was interesting. JRuby, Rubinius look like a lot of fun to hack with. The ruby folks also have a good grasp on systems administration, admiringly. Capistrano and friends assist them in automatically deploying software.

Rails was a bit of a whipping boy at the conference. "Look! Ruby is actually fast! Rails is slow!" As was the MRI interpreter. At the same time there's a lot of love for both. Everyone …