Home Page News: Planning for the Unthinkable

Sep 27, 2007 - 11:25 AM

Working out what to do should a disaster occur used to be complicated, but automated systems have made it easier.

Fortunately, the cost of putting in place effective disaster-recovery and back-up technology is coming down, as this technology migrates down the food-chain — from government and corporates to small- and medium-sized enterprises.

Sam Mulholland, of Dunedin-based Standby Consulting, says the growing dependence on technology is leading to greater use of replication and resilient hardware. Hardware manufacturers now include features like hot swappable disk and power supplies — and software manufacturers include replicating features — in their standard offerings.

Virtual security

The other major trend is a move to virtual machines. This technology is particularly suitable for those with more than 15 servers. Users find virtualisation reliable. It offers excellent recovery, particularly when a whole server can be “dragged and dropped” from one location to another.

Voice-over-IP is another useful technology, as it allows secondary PABXs to step in should a system fail. Load-balancing of call centre work is similarly possible.

Tapes are still used, mainly for off-site storage, explains Mulholland, but tape-based systems can lose data, so replication and mirroring is finding favour. Some clients also split the load between two sites; having two centres allows for upgrades and repairs while keeping systems flowing.

Internet-based solutions are not proving as popular as expected because of unreliability and bandwidth limitations. They might work for small firms, but not for large ones. Some companies, such as Revera, based at Albany on Auckland’s North Shore, are also using storage vaults

Standby Consulting does not itself supply the hardware and software for disaster-recovery projects, as it believes its independence has value in the market. Instead, it advises organisations on what to do.

Mulholland says disasters may only happen once every five to 20 years and some losses are acceptable. But organisations need to reduce their risk, see how they can do this, and analyse the impact of a disaster on their organisation. They then need to work out the recovery sequence of business processes, assessing what is critical and what isn’t.

“Then you can build an effective, cost-effective and cost-justified strategy. Management understands why you need it — the cost; the plans,” he says.

Typically, Standby will devise a business-impact assessment and a DR/BC (disaster recovery/business continuity) plan for customers. This starts with staff interviews regarding what systems are critical. These are then ranked in order of importance and tolerable downtime. Recommendations are then made, which may include a need for new technology.

Health Boards lower their DR risk

Hutt Valley DHB is now installing a storage area network in each of its two computer rooms and virtualising its servers.

“If one server breaks down, you can shift the processing capacity and storage to the second computer-room in real-time,” says CIO Tony Cooke.

The DHB has a BC plan, based on Standby’s templates, and, while it expects localised failures, if there is a major fire it knows where to find the relevant information, says Cooke.

Harry Barber, CIO of Gisborne-based Tairawhiti DHB, also used Standby to assess risks, identify critical components and rank them. Such a “useful piece of work” helped determine spending needs, but left the heath board with a DR/BC plan that is “a living document” which can be constantly be updated.

“DR is only as good as its currency. It’s finding the time to stay on top of it,” Barber says.

The Aviation and Tourism Training organisation also enlisted Standby to help with its planning.

“They identified gaps and weaknesses, and strengths, and suggested how we fix the weaknesses. Some we did, some we didn’t due to cost issues. They simply wrote the plan,” says quality assurance officer Sharon Payne.

Information extracted from Computerworld article of September, 2007

• Back to news