Can OLTP database workloads use Amazon S3 as primary storage? Now they can, thanks to the Cloud Storage Engine (ClouSE), but the question is: how fast?

• ClouSE on “across the street” vs. “across the continent” cloud storage
• ClouSE vs InnoDB
• ClouSE benefits

To answer the question about performance, we decided to run db_STRESS benchmark on a MySQL database in Amazon EC2.  We compared 3 configurations:

• “Across the street storage”: ClouSE with data stored in S3 in the same …

MySQL Cluster partitioning key

I’ll be presenting a webinar covering MySQL Cluster performance on Thursday, July 26. As always, the webinar will be free but you’ll need to register here – you’ll then also receive a link to the charts and a recording of the session after the event.

The replay of this webinar is now available from here.

Here’s the agenda (hoping that I can squeeze it all in!):

• Introduction to MySQL Cluster
• Where does MySQL Cluster fit?
• Benchmarks:
  • ANALYZE TABLE
  …

The other day at PSCE I worked on a customer case of what turned out to be a problem with poor data locality or a data fragmentation problem if you will. I tought that it would make a good article as it was a great demonstration of how badly it can affect MySQL performance. And while the post is mostly around MyISAM tables, the problem is not really specific to any particular storage engine, it can affect a database that runs on InnoDB in a very similar way.

The problem

MyISAM lacks support for clustering keys or even anything remotely similar. Its data file format allows new information to be written anywhere inside a table. Anywhere can be either at the end of a file where it can be simply appended or an empty space somewhere in the middle left after previously deleted row(s). This implies no particular order in which rows are stored unless there are absolutely no …

This week we talk about NULLs and how to find the right data type.

Events
OSCon will be held Monday Jul 16th through Friday Jul 20th and has a data track. We did a podcast about OSCon, including a discount code!

MySQL Connect will be held in San Francisco on Saturday September 29th and Sunday September 30th. This is a technical conference about MySQL, bringing together MySQL engineers at Oracle and the MySQL Community, and will include sessions about the latest MySQL features and roadmaps. The …

Cloud-powered BLOB type provides ACID guarantees and fast direct access to blobs via Web URLs.

Storing unstructured data

Typically unstructured data (such as pictures, media files, documents)

a) Is either stored on the file system, unlike the related with it relational data which is stored in the database. This is well known, “convenient” practice that allows fast access to files but offers no transactional story and no unified data management (for db and filesystem)

b) Or is stored in BLOBs. This ensures transactional consistency and reduces management complexities, but is really bad for performance and scalability.

We took advantage of the cloud, and came up with an upgrade to the BLOB – a solution that combines the benefits of the two.

Weblob data type

Weblob is a new data type that is supported by the Cloud Storage Engine for MySQL ( …

I found time to continue my previous work on group commit for the binary log in MariaDB.

In current code, a (group) commit to InnoDB does not less than three fsync() calls:

1. Once during InnoDB prepare, to make sure we can recover the transaction in InnoDB if we crash after writing it to the binlog.
2. Once after binlog write, to make sure we have the transaction in the binlog before we irrevocably commit it in InnoDB.
3. Once during InnoDB commit, to make sure we no longer need to scan …

This article describes how I tuned MariaDB to give the best write throughput with SSD based storage.

When you have a write-heavy application writing into InnoDB, you will probably experience the InnoDB Checkpoint Blues. The effect manifests as stalls – short periods of time where the troughput falls to zero and I/O activity goes crazy. The phenomenon is well known and described i.e. here. More background about checkpointing can be found …

A few days ago MariaDB, MySQL and Percona all three released new versions of the 5.5 server. So I decided it’s time to run sysbench once more and compare the OLTP performance. The test candidates are:

• MariaDB-5.5.24, using either XtraDB (default) or InnoDB
• MySQL-5.5.25
• Percona Server 5.5.24-26.0

For the benchmarks I used our trusty old pitbull machine which has 24 cpu cores, 24G of RAM and a nice RAID-0 composed of 3 SAS SSD.

The benchmark was sysbench-0.5 multi-table OLTP, using 8 tables with total 10G of data. InnoDB buffer pool was 16G, InnoDB log group capacity 4G (the maximum for MySQL). The mysqld process was constrained to 16 cores, sysbench to the other 8.

The only nonstandard tuning was for I/O scalability: innodb_io_capacity = 20000 and innodb_flush_neighbor_pages = none (for Percona Server and MariaDB/XtraDB). See …

This article describes how I tuned MariaDB to give the best write throughput with SSD based storage.

When you have a write-heavy application writing into InnoDB, you will probably experience the InnoDB Checkpoint Blues. The effect manifests as stalls – short periods of time where the troughput falls to zero and I/O activity goes crazy. The phenomenon is well known and described i.e. here. More background about checkpointing can be found …

With the release of MySQL 5.6 binary log group commit is included, which is a feature focused on improving performance of a server when the binary log is enabled. In short, binary log group commit improve performance by grouping several writes to the binary log instead of writing them one by one, but let me digress a little on how transactions are logged to the binary log before going into the details. Before going into details about the problem and the implementation, let look at what you do to turn it on.

Nothing.

Well... we actually have a few options to tweak it, but nothing required to turn it on. It even works for existing engines since we did not have to extend the handlerton interface to implement the binary log group commit. However, InnoDB has some optimizations to take advantage of the binary log group commit implementation.

binlog_order_commits={0|1}

This is a …