How to use MySQL Cluster and MySQL Replication effectively

Washing machine cluster does one billion transactions per minute

NoSQL may be the buzzword of the day, but MySQL is still an extremely powerful tool, used for a whole range of high-profile web applications. With MySQL Cluster, which recently saw its stable 7.4 release, servers are able to handle an unprecedented number of transactions every second – but what’s the easiest way to integrate with PHP?

A world without replication is unthinkable, and many PHP applications achieve web-scale using MySQL Replication, implementing partitioning, parallel processing, edge computing, clustering, replication and horizontal scale-out on massive multi-core commodity hardware. The MySQL product portfolio covers time-delayed, asynchronous, synchronous, real-time replication of both in-memory and on-disk data. The replication system knows about ACID transaction and reaches up to five-nines (99.999%) availability, including automatic failover and automatic sharding. New NoSQL interfaces round up the offering for those who have reached the limits of integrated caching, parallel and in-memory processing.

Developing for distributed MySQL systems

Today’s MySQL replication features go far beyond lazy primary copy (master-slave). PECL/mysqlnd_ms makes using any cluster of MySQL servers easier. PECL/mysqlnd_ms is a mysqlnd [1] library plugin and works with all PHP MySQL APIs (mysql, mysqli, PDO_MySQL). Consequently, it can be used with any database abstraction layer (Doctrine, Zend, etc.). But before discussing the new client tool, let’s bust some myths by presenting today’s MySQL replication features.

Some say that traditional RDBMSs such as MySQL were designed for the sort of hardware that nowadays runs a washing machine. If true, many of today’s web properties [2] run on washing machines, because they run MySQL. And a small washing machine cluster consisting of eight machines handled 2.46 million write transactions per second [3] in 2011. And, apparently, washing machines can have up to 48-64 cores, because this is roughly the scale-up limit of MySQL today.


Unsere Redaktion empfiehlt:

Relevante Beiträge

Meinungen zu diesem Beitrag

- Gib Deinen Standort ein -
- or -