Tips for designing a resilient WAN

IT executives think a lot about disaster recovery these days. And that’s a good thing. As companies continue to consolidate resources – data centres, facilities, networks – the need for redundancy becomes critical.

All too often, though, in their rush to back up their systems and facilities, executives overlook back-up strategies for their networks.

That’s bad news. In today’s distributed environment, losing network connectivity can short-circuit a company’s ability to function just as effectively as a massive system or data centre outage.

Here are some best practices for designing and developing a resilient WAN:

– Ensure physical redundancy. Make sure you have physical diversity in your cable runs (even if the circuits are provided by different carriers, you might need to double-check that the providers aren’t sharing strands of the same cable).

– Ensure logical redundancy. Check that your IP services have alternate routes. Multihome your IP links or set up a redundant connection to an IP network access point served by multiple IP providers.

– Check for carrier facility and power redundancy. It doesn’t do any good to connect to a carrier point of presence if the switches are down because the power is out. Make sure your service providers have back-up sources (including diesel generators)

– Protect remote offices and workers. Don’t assume everything’s fine just because your site-to-site connectivity is in place. Many times, remote offices lose all functionality if they can’t connect back to centralized data and applications. Look into alternate technologies to provide this connectivity, including dialup, home broadband (Digital Subscriber Line and cable modem) and, increasingly, wireless fidelity. IP virtual private networks also can provide a highly effective, low-cost back-up strategy.

What about voice? Make sure your voice network is backed up as effectively as your data WAN. If cell phones are your backup mechanism, make sure you have an up-to-date directory of numbers. And does every employee at your company have a cell phone? For some organizations (grocery stores and other retail organizations, for instance) equipping every employee with a cell phone is impractical. For outgoing calls, calling cards might be an option (assuming your employees can reach a functioning phone). How do you plan to handle incoming calls?

If you’ve implemented IP telephony, make sure you dedicate the same care to backing up your IP telephony system as you have for your voice and data networks. Be particularly sensitive to power issues.

Test regularly. Not long ago, I asked a group of IT executives how often they performed a full soup-to-nuts test of their back-up plans, including facilities, systems, and networks. The answer? Never.

Bad news.

Yes, it’s difficult. Take the time, and figure out a plan. You won’t be sorry.

Finally, develop an effective communications plan. Figure out how to alert folks in the event of an outage, or even a test. If you need them to change their business processes, how will they know to do so?

Bottom line? When you’re thinking of ways to make your data centres, servers, and systems more redundant, don’t forget the network.

Johnson is president and chief research officer at Nemertes Research, an independent technology research firm. She can be reached