The company has been dubbed Continuent, a name that CEO Eero Teerikorpi said is meant to imply the notion of continuous enterprise.
When the company launched in 2001, it was focused on lower-level IP clustering. Its targets were telecommunications and networking equipment vendors—unsurprisingly parallel to the places where MySQL database use was and is common.
Whats changed over the years is that applications with databases on the back end have proliferated, Teerikorpi said. These applications are being promoted on the Internet, and they need to be up 24-7, 365, making continuous service important.
As well, as the need for high availability has grown, its increasingly come from shops with heterogeneous database mixes, said Donna Jeker, Continuents vice president of product marketing and management.
"We see the same requirements of database availability and scalability exists across the corporate environment, but most corporations are using other databases for these applications," she said.
MySQL is still in the mix, Jeker said, but not for mission-critical uses such as transactions.
"They may have MySQL in use somewhere, but not necessarily for revenue-generating purposes," she said.
For mission-critical purposes, most clients Continuent sees are using PostgreSQL, she said—for "when you dont want to lose a transaction."
Thus, offering clustering for the open-source PostgreSQL database makes sense, she said, although the company isnt stopping with just open-source databases.
"…Where were going with production direction [is] to support any major database with middleware functionality and also look at commercial databases," she said.
Of course, the database vendors themselves have come up with high-availability offerings.
Oracle, for example, has RAC (Real Application Clusters). Microsoft, on the other hand, dropped mirroring off its feature list for the upcoming SQL Server 2005, due out next week.
Regardless, these are point solutions, Teerikorpi said, that only support specific databases.
Oracle RAC, for its part, uses a shared-storage approach, which is a "pretty expensive back end for data storage," he said.
Continuents products differ in that they use shared-nothing architecture. Each node supports its own, local database. Not only do nodes support clusters, they can also run in parallel, using the same application.
That means that to scale out, a mix of databases can be used, including the open-source databases MySQL and PostgreSQL—a very cheap alternative to scaling out with proprietary databases.
Continuent claims that its newly launched database virtualization technology is the first virtualization middleware that supports all major commercial and open-source databases and application servers.
The companys m/cluster for MySQL and p/cluster for PostgreSQL products provide transparent failover and disaster recovery for transactions.
They also provide automatic continuity at the transaction level with no lost transactions.
Continuents product line allows a number of configurations to be installed that provide vertical scale-out: the ability to add low-cost commodity hardware while maintaining the image of a single large database to the application tiers.
It also supports data partitioning, giving additional scalability, and has load-balancing capabilities.
Continuent runs on commodity hardware and Linux. It features a single graphical management console that provides a global view of the virtual databases and the database cluster.
Earlier in the month, Continuent.org, an open-source portal and community dedicated to high availability and scalability for databases and related technologies, was launched. Continuent sponsored the portal.
Continuent m/cluster for MySQL is available now starting at $4,995 per database CPU. p/cluster for PostgreSQL, uni/cluster for universal database support, ms/cluster for SQL Server, s/cluster for Sybase and o/cluster for Oracle will be available in 2006.
The expanded product line is being funded by $5.75 million in a series B round of funding.
Jeker said the purpose of the funding is to generate the new product infrastructure, which will focus on heterogeneous database support—both in normal operations and for database migrations, she said.