With a few changes to InnoDB and MySQL I think I can get more than 100,000 disk reads/second from MySQL 5.1.47. InnoDB and MySQL continue to amaze me. This is a remarkable result.

Storage devices that support more than 100,000 IOPs can be purchased for commodity servers. This capacity is much more interesting when InnoDB can use all of it or when storage vendors give you a discount because your software can't use their capacity. My first tests were done with an 8-core server and I was able to get ~12,000 disk reads/second from InnoDB in MySQL 5.0.84 and ~18,000 reads/second from the InnoDB plugin in MySQL 5.1.47. My second tests were done on a 16-core server and I was able to get ~18,000 reads/second from 5.0.84, ~30,000 reads/second from …

MySQL does not follow the ANSI SQL standard for quoting. MySQL’s default quoting behavior is that either single or double quotes can be used to quote a string (this gets me into trouble when I work with Oracle databases, as double quotes do not indicate a string!).

mysql> SELECT 'alive';
+-------+
| alive |
+-------+
| alive |
+-------+
1 row in set (0.00 sec)

mysql> SELECT "alive";
+-------+
| alive |
+-------+
| alive |
+-------+
1 row in set (0.00 sec)

Bare words are dealt with in context; in this case, a bare word would be parsed as a column name:

mysql> SELECT alive;
ERROR 1054 (42S22): Unknown column 'alive' in 'field list'

Backquotes are the way MySQL escapes table names. So, if you want a reserved word, number or operator to be the name of an object (ie, a table named “1″ or a column named “date”) you need to use backquotes to avoid a syntax error….for example:

mysql> …

It’s Friday, and summer’s here. While it seems the industry is slowing down to a lazy pace, there is still some action so let’s splash right in to this week’s edition of Log Buffer DBA industry news in Log Buffer #192.

Alex Gorbachev had a few minutes to suggest the following interesting tidbits to me before running off to attend Oracle ACE Director activities at ODTUG/Kaleidoscope this weekend. One of these days we’ll have to see if he can share some of what goes on behind closed doors at those hush …

Apache And MySQL Monitoring With Bijk On Debian Lenny

This tutorial describes how you can monitor your server with the tool Bijk. Bijk creates online 30 graphs about load, CPU, memory, traffic, Apache, NginX, PostreSQL and others with alerts. Bijk can be used on Debian, Ubuntu, CentOS, RedHat and Gentoo. In this article I will explain how to install Bijk on Debian.

It is time… Time for one more very special conference for me. Why special?

1. I have never been to ODTUG Kaleidoscope before. I always like new conferences — new experience.
2. Unlike Sheeri, I do not speak! This is one of those rare conferences where I come to slack off, meet old friends and make new ones, go to lots of sessions and actually learn stuff.
3. Coming from the DBA background, it’s not often that I come to development oriented conferences and I think I should do more of that.
4. It’s in Washington, DC. I lived there for some time and have number of good friends there. I’m really excited to see them again!

So what am I going to do there? I just arrived and right in time for the Oracle ACE Directors’ briefing that will run for the whole days of …

In celebration of Midsummer today, I wanted to post the below monologue on likelihoods (of nuclear powerplant catastrophies), which is a translation of a famous Swedish monologue by Tage Danielsson.

When we at MySQL had joined Sun, one task for me and my Sales Engineer collagues was to travel to Sun offices and educated the huge Sun sales force about MySQL, so they could sell it too. (Basically to tell them about open source, scale-out, reference customers, and most importantly: Don't sell Cluster on your own, call me first.) Being a Telecom Sales Engineer, I was sent to tour the Ericsson account team meeting, the Nokia account team meeting, and for logistical reasons even the Siemens account team meeting that was at the same location as the Nokia team.

Each meeting had nicer and nicer dinners, but the Ericsson account team meeting in Stockholm was clearly the winner. The dinner was set in the City Hall restaurant …

1. Membase -- an open-source (Apache 2.0 license) distributed, key-value database management system optimized for storing data behind interactive web applications. These applications must service many concurrent users; creating, storing, retrieving, aggregating, manipulating and presenting data in real-time. Supporting these requirements, membase processes data operations with quasi-deterministic low latency and high sustained throughput. (via Hacker News)
2. Sergey's Search (Wired) -- Sergey Brin, one of the Google founders, learned he had a gene allele that gave him much higher odds of getting Parkinson's. His response has been to help medical research, both with money and through 23andme. …

I tried the Linux instructions on my Mac, and it worked without need for any changes. Things may be different when you deploy a real set of servers on separate hosts, but it's a good …

Single host Cluster

A series of quick-start guides are now available to get you up and running with MySQL Cluster in as little time as possible; they are available for LINUX/Mac OS X, Windows and Solaris. The configuration is intentionally a simple one – 2 data nodes, 1 management node and 1 MySQL Server. Once you have this up and running, your next experiment may be to extend this over multiple hosts.

Download the Quick Start Guide for your platform below:

• MySQL Cluster Quick Start for Linux – also works for Mac OS X
• …

So running through some various tasks, I'm finally on the section of work where I can resurrect a script that finds inconsistent data between master-master pairs.

Let's get a quick summary to find our problems to target the script at.

dsh -ef 20 -w dbfacebook9b,dbfacebook9a -s ~/bin/execute_query.sh "FacebookShard \"CHECKSUM TABLE RollingUserLog\""
dbfacebook9b: *************************** 1. row ***************************
dbfacebook9b:    Table: FacebookShard.RollingUserLog
dbfacebook9b: Checksum: 538386033
dbfacebook9a: *************************** 1. row ***************************
dbfacebook9a:    Table: FacebookShard.RollingUserLog
dbfacebook9a: Checksum: 538386033

not bad.

CHECKSUM TABLE uses a ACCUM algorithm to determine the table checksum. This gives a good fast snapshot in a shared lock mode so writes do not stop while doing a …