Symantec makes high-availability server clusters more application-aware |
 |
By Beth Pariseau, Senior News Writer
08 Oct 2008 | SearchDisasterRecovery.com |
|
|
Symantec Corp. has revamped its Veritas Cluster Server software with a new release called VCS One aimed at improving high availability recovery of entire business applications.
The updated software is a lighter weight agent than previous hardware-centric versions of high-availability (HA) clustering, Symantec claims. The aim is to allow organizations to keep active farms of virtual servers running at a disaster recovery site, as well as recover tiered applications with dependencies intact.
Mark Lohmeyer, vice president and general manager of the VCS product group at Symantec, said previous versions of the server failover software were "tightly coupled" with their underlying hardware. Customers needed to maintain a strict match between source and target hardware, as well as operating systems.
VCS One, which became available this week, includes a clustering agent that runs in user space outside the operating system kernel. That means applications running on multiple server clusters can communicate with one another, even if they run in multiple sites or have firewalls between them. Applications can also be recovered between disparate physical and virtual servers, and multiple operating systems can now share the same cluster. However, source and target servers must be running the same operating system for recovery to work.
A central-policy master server monitors each server agent. The master server can initiate failover and recovery according to user-set policies. With this release, the master server can support 256 mixed operating system nodes at two sites.
"Because we're able to exist across all those servers and capture dependencies between them, we can automate the restart of an entire application service," Lohmeyer said. The application service might consist of several layers, including database, application and Web servers. "Right now, this process is dependent on a lot of tribal knowledge in the heads of individuals who know the right order and design scripts to run this kind of recovery," he added. "We're looking to get people out of that process and lower the risk of error."
Customers can maintain pools of computing resources at a disaster recovery site that can be used to run test and development applications, or share the application workload with the primary site. Those resources can then be shifted in the event they're needed for a recovery. "You don't need to have server resources waiting idly for a disaster," Lohmeyer said.
Robert Ball, director of services and solutions for value-added reseller (VAR) ServerWare, said it's usually his practice with clients to retain 50% overhead on all servers in order to leave room for failover if it's needed. But a client might, for example, maintain a four-node cluster for the primary environment and a two-node cluster for test and development at another site. "VCS will automatically shut down the applications on the secondary cluster if the first site needs to failover," he said.
VCS One will ship with preset policies for common tiered applications, such as Oracle, SAP, PeopleSoft and DB2. Agent development kits are available for customers who want to support another application with VCS One. Pricing for VCS One is $995 per CPU on protected servers. There is no charge for the policy master server.
Today, VCS One supports Windows, Linux and Solaris operating systems, and virtual servers VMware, Linux Logical Domains and Solaris containers.
Symantec isn't offering Xen support with this release despite a partnership with Citrix for server virtualization, but Lohmeyer said VCS One was under development before Xen came on the scene. "[Support for Xen] will be in our very next release," he said.
Ball still hopes for the ability to eventually support failover between multiple operating systems, as well as between different physical and virtual hardware resources. "We can fail file systems, for example, between Solaris and AIX boxes, but the [application] binaries are still different," he said. "The dream would be to failover an application from a Windows server, run it on Linux and have it be supported."
|
|
|
|
