To be launched at the MySQL Users Conference & Expo in Orlando, Fla., MySQL Cluster combines the MySQL database with a clustering architecture for 99.999 percent availability for mission-critical applications, according to the company. The database features a distributed in-memory clustering architecture to boost availability and throughput, with response times of five milliseconds to 10 milliseconds and a throughput of 100,000 replicated transactions per second on a four-node cluster with two CPUs per node.MySQL Cluster will be offered under MySQL's dual license business model, in which it is provided at no cost under the open source GNU General Public Linense. The commercial version if priced at less than $5,000 per processor they said. Under the GPL license, users must publish source code for their applications.
“For the application developer, [failover] is automatic. They don’t have to worry about underlying failure detection algorithms or how the database is distributed across multiple nodes,” said Zack Urlocker, MySQL vice president of marketing.
Data can be distributed over a group of interconnected databases on multiple servers or nodes. Companies have been testing the database in a 48-node configuration, according to MySQL.
Although the clustering product offers linear scalability in line with competitive products from vendors such as Oracle and Microsoft, the company with the clustering product is attempting to compete against established vendors such as Oracle or Microsoft, Urlocker said. “We’re responding to clustering [needs] for MySQL customers,” he said.
Oracle, for example, has advanced grid capabilities in its database and MySQL has no intention to focus on grid, Urlocker said. The goal is to make clustering mainstream with MySQL Cluster, he said.
It will run on Linux, Windows, Solaris, Mac Os X and others. It supports various hardware platforms like 32 and 64bit Intel systems, PowerPC and Sparc.
Also announced was a partner program to boost compatibility with MySQL.