The latest from the MySQL community

I’m real excited about our upcoming seminars for Oracle DBAs who want to learn MySQL. Having used Oracle myself for many years, I know full well there are lots of cases where Oracle and MySQL are a great combination to have in a side-by-side manner. There are things you can do with Oracle that you can’t do with MySQL and (believe it or not…) there are things you can do with MySQL that you can’t do with Oracle. Plus we have lots of folks who offload Oracle application data to MySQL data marts/warehouses or other systems. This being the case, it’s not surprising that even the IOUG (International Oracle Users Group) admit that now 1/3 of Oracle shops use MySQL.

The upcoming seminars, scheduled for New York (August 14th) and San Francisco (August 16th) don’t contain fluffy presenations, but instead focus on the hard-core tech details DBAs love. Oracle DBAs will leave these seminars armed with the information they need to smartly …

One of the major contributing factors to the success of MySQL is the ease and simplicity of its replication. Read-slaves for scaleout and backup-slaves for noninterrupting backups are the norm in any MySQL installation I have seen in the last two years.

So how does replication work? And how shall it be expanded in the future?

What is available?

The binlog written by MySQL currently logs all statements changing the tablespace. It is a serialization of all tablespace changes. The binlog position, expressed as (binlog name, offset), is a database global timestamp - a timestamp expressed in seconds.fraction does not work for any precision at all, because on a multi-core machine multiple things can happen concurrently.

If you want to make a consistent full backup of the database, the database must not change during the backup. That is, it must be possible to associate one and exactly …

OpenMoko releases open source phone. Turbolinux works with Microsoft on Open XML. Alfresco releases updated Community Edition. (and more)

Openmoko goes live, Openmoko (Press Release)

Major Linux Distributor Joins Open Source Open XML-ODF Translator Project, Turbolinux / Microsoft (Press Release)

Alfresco Innovates with New Community Edition, Delivers First-Ever Open Source Mashable Enterprise Content Services, Alfresco Software (Press Release)

New Python-based database communication tool represents first open source …

I test updates, queries and their mixed cases. It works well.
I haven't measure the performance impact yet. I think the influence of Aspect C should be small since it just adds function calls. If the inline is supported when compiling, the cost should be minimum.

Test 1:
./sysbench --test=oltp --oltp-table-size=1000000 --mysql-table-engine=pbxt
--mysql-engine-trx=yes --mysql-user=root --mysql-socket=/tmp/mysql.sock
prepare --mysql-db=test

SysBench Result 1:
sysbench v0.4.8: multi-threaded system evaluation benchmark

No DB drivers specified, using mysql
Creating table 'sbtest'...
Creating 1000000 records in table 'sbtest'...

Profile Result 1:
System_call Total_Execution_Time Total_Call_Times
pread 719098 24312
pwrite 11799545 1026503
malloc 732 453
calloc 0 0
realloc 0 0
pthread_mutex_lock …

I was just reading this article on PC World about the launch of GPLv3. I reckon it's a decent article in that it actually bothers to explain the foundation of GPL (Richard Stallman's software freedom principles).

One quote (from an engineering professor at MIT) gets it wrong though:When something in the free software world gets improved, there's no waiting to buy a new license, everything is shared with all users, so everyone benefits[...]That's just not true. GPL does not invoke an obligation to publish modifications. Of course, when the resulting work is distributed, the source code should also be made available to those same people. But that's not quite the same. And there's more to it. Let's take the following example of how you could do business (that is, make money) …

I've written the following program to profile system calls by using Aspect C.
Put this file (profile.acc) inside the src directory, ACC will automatically weave it with the original code, and then compile it.

#include
#include "xt_defs.h"

#undef PROFILE_DEBUG

/*
* profile_get_time
* Return: current time in micro seconds
*/
inline long profile_get_time( )
{
struct timeval current_time;
gettimeofday(&current_time,0);
return (current_time.tv_sec)*1000000+current_time.tv_usec;
}

/*
* profile i/o function: pread
* we record its execution time when pread is called
*/

long profile_pread_access_time=0;
long profile_pread_freq=0;

int around(): call($ pread(...)) {

long start_time = profile_get_time();

int val = proceed(); …

Tonight I spoke at the NY MySQL Meetup . The topic “Practical Performance Tips & Tricks” was a full packed 1 hour session, with 4 x 15 minute sections on Beginner, Intermediate, Advanced and MySQL Proxy.

I haven't been blogging about the things I used to -- how-tos and technical hacks -- because I've been working quite hard on MySQL Toolkit and, believe it or not, innotop. I've made it possible to write innotop plugins, which have been very useful to our team at work, and I'm working on documentation. Plugins won't make it into the upcoming release; it'll just be bug fixes and documentation. These projects have taken up most of my free time.

I've also tentatively joined a group of people working on a very large, exciting, secret project which may consume the rest of my summer and/or much of the rest of the year. (I've been turning down consulting gigs and other projects because of this). If this secret project works out, I'll be writing more about it here, you can count on that.

In the …

This release fixes some bugs and improves the chunking functionality. MySQL Table Checksum had a few minor bugs and one major bug with the chunking functionality. I also rewrote the chunking, though the behavior is backwards compatible. I am very happy with the way it works now, and will probably not make any more incompatible changes to it. The changes enabled me to add support for chunking on float, double and decimal columns.

It still doesn't support chunking on character-based columns, though I know now how I'll do it if I do. Also, support for ENUM and SET shouldn't be hard to add. I have no need for these features myself. If you need it, please file a bug report on the Sourceforge tracker.

Ask a simple question, get a simple, but subtle answer. I asked Savio Rodrigues, who replaced me on the Open Sources blog but originally blogged here, to comment on the state of open source at IBM. He gave me a bit more than that.

You know, IBM, the company that essentially carried open source into the enterprise on its back in 2000 when it pledged $1 billion to fund Linux. Lately, though, IBM's has been less flashy with its commitment to open source though, as Savio points out, no less involved. As Savio reports, however, IBM's commitment to open source is broader than source code. Open source without open standards isn't of much interest to Big Blue.

In this fifth installment of the Open Source @ Series on The Open …