Veritas CFS Proves that Fast Failover Minus High Costs, Complexity and Uncertainty is Attainable

The tolerance for downtime in enterprises is nearly gone. Business requirements have changed, service levels are more stringent and every second of an outage could translate into millions of dollars in lost revenue. Classic high-availability (HA) solutions may no longer cut it as enterprise applications require fast failovers, often with sub-minute recovery times.
 
Yet this level of HA generally comes with a steep price tag and extra overhead that can preclude organizations from deploying any applications other than their most mission-critical in this fashion - typically transaction-oriented or database applications. Veritas Cluster Server (VCS) combined with Veritas Storage Foundation Cluster File System (CFS) delivers faster failover without the added cost and complexity. 

In the past, classic HA solutions have worked well, but each has had its own limitations:

• Classic HA solutions are set up as active and standby servers which results in a doubling of operational costs due to the hardware requirements.
• Native Operating System (OS) HA solutions are available but the real costs rear their heads in licenses, maintenance and configuration costs. In addition, customizing the HA solution for other applications that are running in the environment is no simple task.
• Application-specific solutions fragment the process for managing the availability in the environment which increases management complexity.  If other applications are running in these environments they will require their own HA solution. In most cases, these application-specific solutions do not address OS, hardware or network failures within the environment.

Further complicating the situation, these classic HA solutions may not failover in the sub-minute time frames that organizations expect.  In a classic HA environment that uses an active/stand-by configuration, when the active server fails, the application failover will take three steps:

1. Detect the failure.
2. Prepare the standby server to recover the application and migrate data. This could take upwards of a half an hour to fail for a really large database and, in a worst case scenario, it could take days if there is data corruption in the file system.
3. Restart the application on the standby server   

In order to address these performance issues, organizations can go the active/active route: Oracle RAC. However, the software licensing costs for Oracle RAC implementations can be onerous. Aside from the hardware costs (the minimum number of servers to serve an Oracle database in an Oracle RAC configuration is two), organizations need to pay for Oracle licensing on each server.  In this scenario, up to 70% of the costs of an Oracle RAC environment could easily come from Oracle licensing and yearly maintenance.

This is not to say that there are no use cases for these clustered solutions. Assuming they are appropriately configured and all of the backend resources are in place, they can achieve rapid failover times (possibly in 5 - 10 seconds or less). The downside is that organizations will pay in terms of the hardware and software that is necessary to implement and support it.

This is where Veritas Cluster Server (VCS) combined with Veritas Cluster File System (CFS) differentiates itself as an all-purpose, fast failover solution for any application on any UNIX and Linux platform:

• It creates an HA configuration where should any production server fail, failover occurs on the standby server (or anywhere there is capacity) automatically without any administrative intervention.
• It eliminates the step of bringing storage online at the standby server, typically the longest part of the recovery process.  This is because each server has access to the same storage already.  This means that recovery times are now just the time it takes to restart the application or database.
• It eliminates the complexity and uncertainty that is typically associated with the setup and management of CFS environments. Veritas Cluster File System sits atop Veritas Storage Foundation so Storage Foundation abstracts much of the complexity associated with managing the underlying block storage that other CFS solutions first have to deal with.
• It eliminates many of the hardware and software licensing costs associated with OS and database clustering solutions. Symantec HA solutions can reduce the ratio of production application servers to standby servers to a ratio an organization feels comfortable supporting.
  

So, assuming an organization finds a 7:1 ratio of production servers to standby servers appropriate for its environment, with Symantec organizations can drop the number of physical servers solely dedicated as idle failover servers for each application from 7 down to 1 plus they do not need to pay a fee to license the Oracle database on the standby server.

• While Veritas CFS has a licensing fee, it is generally less expensive than comparable application and database clustering solutions. In addition, since less standby servers are needed for HA, fewer software licenses are needed. This reduces an organization's upfront software costs as well as its annual software maintenance costs.

Complexity, high costs, manpower and uncertainty are not necessarily synonymous with sub-minute application recovery times if the right solution is deployed. By leveraging Symantec's Veritas Cluster Server and Cluster File System, organizations eliminate the complexity and uncertainty associated with deploying sub-minute recovery solutions while also needing less hardware and software to deliver it. In doing so, Symantec delivers the type of affordable and hassle-free solution that organizations in today's age of always-on computing want and need.