The MySQL 5.6 Release has introduced some changes to how two phase commit works and is managed.  In particular, the commit phase of transactions to the binary log is now serialized and this behavior is something we identified fairly immediately.  We implement a group commit algorithm that needed to be altered so that TokuDB’s group commit to its recovery log would function effectively.

As part of our effort to verify the new Binary Log Group Commit functionality introduced in TokuDB 7.5.4 for Percona Server, we wanted to demonstrate the substantial increase in throughput scaling but also show the bottleneck caused by the skewed interaction between the binary log group commit algorithm in MySQL 5.6 and the transaction commit mechanism used in TokuDB 7.5.3 for Percona Server.  During our testing, we noticed that the throughput scaling was diminished when we turned on the binlog.

Here are the relevant system …

Facebook recently made opensource, osquery. It gives you operating system data via SQL queries! Its very neat, and you can test this even on MacOSX (it works on that platform & Linux). It is by far the project with the most advanced functionality, linked here in this post.

I noticed that rather quickly, there was a PostgreSQL project, called pgosquery, based on Foreign Data Wrappers with a similar idea. (apparently it was written in less than 15 minutes; so a much lower learning curve than the regular MySQL storage engine interface)

I immediately thought about an older MySQL project, by Chip Turner (then at Google, now at Facebook), called …

Today we released TokuDB® v7.5, the latest version of Tokutek’s storage engine for MySQL and MariaDB.

I’ll be publishing two blogs next week to go into more details about our new “Read Free Replication”, but here are high level descriptions of the most important new features.

Read Free Replication

TokuDB replication slaves can now be configured to process the binary logs with virtually no read IO. This is accomplished via two new server parameters: one to allow the skipping of uniqueness checks (for inserts and updates), the other to eliminate read-before-write behavior (for updates and deletes). The two other conditions are that the slave must be in read-only mode and replication must be row based.

Hot Backup Now Supports Multiple Directories (Enterprise Edition)

The original implementation of our Hot Backup functionality was only capable of …

Covering indexes can result in orders of magnitude performance improvements for queries. Bradley’s presentation on covering indexes describes what a covering index is, how it can effect performance, and why it works. However, the definition of a covering index can get cumbersome since MySQL limits the number of columns in a key to 16 (32 on MariaDB).

Tokutek introduced multiple clustering indexes into MySQL to address these problems. Zardosht describes the multiple clustering indexes feature and how clustering indexes differ from covering indexes. Zardosht also describes the …

June 17, 2014 By Severalnines

We continuously see great interest in MySQL load balancing and HAProxy, so we thought it was about time we organised a live webinar on the topic!

As most of your will know, database clusters and load balancing go hand in hand. 

Once your data is distributed and replicated across multiple database nodes, a load balancing mechanism helps distribute database requests, and gives applications a single database endpoint to connect to. 

Instance failures or maintenance operations like node additions/removals, reconfigurations or version upgrades can be masked behind a load balancer. This provides an efficient way of isolating changes in the database layer from the rest of the …

During April’s Percona Live MySQL Conference and Expo 2014, I attended a talk on MySQL 5.7 performance an scalability given by Dimitri Kravtchuk, the Oracle MySQL benchmark specialist. He mentioned at some point that the InnoDB double write buffer was a real performance killer. For the ones that don’t know what the innodb double write buffer is, it is a disk buffer were pages are written before being written to the actual data file. Upon restart, pages in the double write buffer are rewritten to their data files if complete. This is to avoid data file corruption with half written pages. I knew it has an impact on performance, on ZFS since it is transactional I always disable it, but I never realized how important the performance impact could be. Back from PLMCE, a friend had dropped home a Dell R320 server, asking …

In recent time, MariaDB 10 has been getting many new storage engines. We’ve seen TokuDB, CONNECT, SEQUENCE, SPIDER, CassandraSE for various use cases. For a long time, MariaDB shipped OQGRAPH, but it was disabled in MariaDB 5.5. It will make a come back as OQGRAPH v3 has been worked on actively by Andrew McDonnell. Keep track of this via MDEV-5319.

Another engine being worked on by Kentoku Shiba & team is the mroonga engine, which allows you to do full text search. It is optimised for CJK languages, and is supposedly very fast. To track this, follow …

What does it mean if MySQL returns the ‘Incorrect key file for table’ error for one of my queries? The answer is complicated and depends on which storage engine is returning the error. We have debugged two cases which we describe here.

File system out of space

When running the random query generator, one of the queries failed.

Query: SELECT * FROM (mysql . general_log AS table1 INNER JOIN INFORMATION_SCHEMA . INNODB_BUFFER_PAGE AS table2 
ON ( table2 . SPACE = table1 . user_host ) ) ORDER BY table1 . thread_id LIMIT 168 
failed: 126 Incorrect key file for table '/data/mysql7/performance_schema_vardir/tmp/#sql_6b8_17.MYI'; 
try to repair it

Since this query requires a sort, MySQL creates a hidden temporary table called ‘#sql_6b8_17.MYI’ to hold the intermediate results. While the query was executing, some operation performed on the MyISAM table returned an error. What could it be?

MySQL maps the …

Today before Ivan’s tutorial, he told me that in the 10.0.5 virtual machine images he created, he couldn’t find the Cassandra storage engine. I told him it had to be installed separately, and this is true – you have to install some engines separately!

When you do a yum install MariaDB-server MariaDB-client like the installation instructions tell you to do, you don’t get all storage engines (so running SHOW ENGINES might have you wondering what happened to a bunch of engines). This can easily be seen by doing a yum search MariaDB. On a CentOS 6.4 server with the MariaDB 10.0 repository configured, you should see the following:

MariaDB-cassandra-engine.x86_64 : MariaDB: a very fast and robust SQL database server
MariaDB-client.x86_64 : MariaDB: a very fast and robust SQL database server
MariaDB-common.x86_64 : MariaDB: a very fast and robust SQL …

While integrating TokuDB into MySQL 5.6, we found that MySQL 5.6 does not support more than one XA storage engine. For example, there is an assert in the ha_recover function that fires when the total number of XA storage engines is greater than one. After disabling this assert, we found lots of bugs in the MySQL 5.6 implementation of the TC_LOG_MMAP class, which is used when running with the binlog turned off.

There are two alternatives that we know of to fix this problem in MySQL 5.6:

• First, we could merge code from MariaDB 5.5 into MySQL 5.6. The advantage of this approach is that we have been running this code with TokuDB in MariaDB 5.5 for a long time, so we have confidence in its correctness.
• Second, we found that MySQL 5.7.2 has made changes to allow multiple XA storage engines. This is great news for TokuDB since we have one less MySQL patch to worry about. Our simple scan of the MySQL 5.7 source showed …