The Linux Foundation has formed a new working group to speed development within the Linux ecosystem that would make the operating system kernel more suitable for building high availability (HA) systems, the Foundation announced Wednesday.
The High Availability Working Group will define a software stack for running Linux in clustered, mission critical environments. It will also prioritise development work that still needs to be done, based on feedback from developers, vendors and customers.
Engineers from Novell, Oracle and Red Hat, among other companies, will participate in the working group, as well as leaders from Linux distributions such as Debian, Fedora, OpenSuse and Ubuntu.
"The users [of the stack] would be everyone who wants to run a Linux cluster on open source software," said Lars Marowsky-Bree, a Novell engineer who is the working group's architect. While HP offers a HA package for Linux, and Red Hat offers HA capability in a preview mode for Red Hat Enterprise Linux, users have been requesting a fully open source and vendor-neutral package, Marowsky-Bree explained.
The Linux Foundation defines high availability computing as systems that maintain at least 99.999 percent (or "five nines") uptime. This translates to about five minutes and 26 seconds of downtime per year.
Over the past few years, HA system builders have tended to use commodity hardware, with extra redundant equipment in place to ensure the systems keep running even when individual components fail.
By clustering the servers, "they achieve a higher availability than any single one of them would achieve," Marowsky-Bree said.
So, not surprisingly, the Linux HA stack will include a lot of components that should aid in the clustering of servers. In addition to Linux, the software stack may include technologies such as the Corosync cluster engine, the Open Clustering Framework, the Linux Virtual Server, the Pacemaker resource manager, the Oracle Cluster File System (OCFS), the Global File System (GFS) and others.
While many of these technologies are already mature, additional work remains to be done in assuring they all work together easily, Marowsky-Bree said.
Find your next job with techworld jobs