Symantec takes new tack toward clustering

By Kevin Komiega

-- Virtual server technologies are forcing storage vendors to rethink and redesign products on what seems like a daily basis. However, making storage systems and architectures designed for a physical world work in the ethereal world of virtual machines sometimes causes things to break.

Armed with that knowledge, Symantec has gone back to the drawing board and developed Veritas Cluster Server One (VCS One), a new high availability and disaster-recovery clustering platform that turns traditional clustering on its head and offers active-active DR, multi-tier business application HA, and scale-out control for complex IT environments.

According to Rob Soderbery, senior vice president of Symantec's Storage and Availability Management Group, traditional clustering architectures are typically optimized to use between two and sixteen nodes and are built to protect databases. Now, with the advent of server virtualization and its ability to accommodate multiple operating systems in one software stack, the clusters of yesterday are having trouble meeting the needs of today.

"The combination of virtualization and the move to more scale-out computing and storage architectures is breaking traditional clustering systems," says Soderbery. "Traditional clustering solutions are also kernel-resident. How do you keep track of whether you are in the kernel of a virtual or physical machine?"

To combat the shortcomings of older clustering systems, Symantec redesigned its clustering architecture using a client-server approach that scales up to 256 nodes and has server virtualization abstraction and control to deliver a common HA/DR operating model across the various virtual platforms in a data center.

The VCS One architecture operates outside the operating system kernel to facilitate technology deployments and upgrades without downtime. VCS One is also policy-driven, with priorities, scheduling, and dependency management to automate tasks.

"VCS One takes a lot of the clustering infrastructure out of the host and puts it in a master server, which makes it easier to manage physical and virtual servers across different operating systems with more detailed levels of application control," says Soderbery.

Symantec VCS One features an active-active disaster-recovery configuration that makes use of all available gear at a DR site to ensure high- priority workloads keep running regardless of the outage scenario. Soderbery says this approach cuts down on costs by eliminating the need for dedicated spare and idle capacity at secondary sites set aside to handle failures at a primary data center.

For multi-tier high availability, the VCS One platform contains policies for application dependencies that automatically coordinate restart relationships between distributed application components running on various operating systems.

Veritas Cluster Server One is priced at approximately $995 per CPU.

Related articles:
Symantec adds de-dupe, CDP to NetBackup
Symantec enters online backup with SaaS
Symantec updates Storage Foundation

This article was originally published on October 08, 2008