In the grand scheme of business continuity planning, high-availability (HA) systems can play a small but vital role in resuming business activities in the event of smaller -- but still critical --s ystem failures. Specifically, HA can help prevent difficult and inconvenient rerouting issues and other failover snags, problems often encountered when relying on remote-availability solutions.
In cases where cost and failover time are important issues, you must create failover scenarios for your critical data systems that won't break the bank and won't have long failover times. The answer is an HA solution, which offers the ability to fail over locally to readily available and preconfigured systems designed to stand in for the failed system.
HA systems offer several advantages. You can generally fail over very quickly, using 'hot-standby' servers that don't require powering up, rebooting, or configuration. This also means that you don't have to reroute end users because the failover systems are usually on the same IP segment as the primaries.
However, when implementing an HA system, you must be aware of two major drawbacks. Fortunately, you can easily overcome both issues with the proper planning.
HA systems share the first drawback with local disaster recovery solutions. A loss of the physical site could result in the destruction of the backup systems as well as the production systems.
You can avoid this by either replicating data off-site (to much fewer servers and with much less infrastructure) and/or by removing tape backup media to off-site storage locations. Either of these steps can provide the ability to recover from both small and data-centre-level disasters, albeit with different recovery point objectives.
The second drawback is that failover solutions of any kind -- both local and remote -- are much more complex than simple DR solutions. You must configure additional server systems to stand in for one or more production systems, keeping in mind that some systems will not support many-to-one failover.
While this drawback is not necessarily difficult to overcome, you can't overlook it. Make sure to prepare for the implementation with training, professional services, and other methods of getting over the learning curve.
HA systems allow organisations to fail over much faster than remote failover, and they offer a greater level of protection for critical systems than DR alone. If you can overcome the complexity issues involved in the implementation -- and still plan effectively for site-wide disasters -- you can create a solution that meets the needs of your organisation and the bottom line as well.
Talkback
High Availability
Applications which on a day-to-day basis must have little-or-no down time, must be provisioned robustly within the data center to provide a Service Level said to provide “high availability” for the system (e.g., Availability: 99.9%-99.999%, Response Time: 1-4 seconds, Transaction Volume 99.9%-100%).
A robust technical design within a data center for an application which provides high availability, however, does not provide a viable disaster recovery capability for the application, as a disaster recovery solution must be able to recover the system within an acceptable amount of time (as set by the supported Business Units)even in the event of a total data center loss.
Disaster Recovery Vs High Availability
Disaster recovery planning and provisioning for an application, ensures that regardless of what disaster befalls an application, its data center, or its infrastructure, the application can be recovered to acceptable functionality within an acceptable amount of time as defined by its supporting business unit(s). Thus, disaster recovery planning and provisioning must utilize a remote location for recovering the application.
Disaster recovery solutions can be expensive. Generally speaking the quicker the recovery time objective (RTO), the more expensive the disaster recovery provisioning will be. For that reason, an application/system disaster recovery time must be set in terms of the maximum outage that the company can afford before severe, crippling, or long term adverse impacts are incurred by the company. To set the RTO quicker will often result in the cost of provisioning a disaster recovery solution to meet that RTO being prohibitive, or exceeding the risk being mitigated.
Last, where a system is so critical to the survival of the company that virtually no down time under any circumstances is acceptable, lest the company potentially fail, high availability disaster recovery provisioning (split processing, mirroring) becomes a part of the overall high availability solution for the system.