Proper Planning for Network Reliability in Data Centers

"Network Reliability" and "Downtime" are the quintessential business continuity keywords. There are a number of issues that must be considered when creating or maintaining a network to ensure it is reliable. Below is an excerpt from Prepare for the Worst, Plan for the Best: Disaster Preparedness and Recovery for Small Businesses by Donna R. Childs, which discusses initial tests and procedures to ensure you have a reliable network

Network failures, especially for local area networks, are relatively rare. But a network is a good first test on how far contingency planning has progressed. When someone tells you that their network is completely protected and fail-safe, tell them that you would go over to their office and 'pull the plug' on any one network cable. Observe their response!

Before you see network failures, you will typically first face network performance issues, like overloads of routers or switches, which clearly illustrate the need for good network capacity planning especially if your network exceeds 100 users. If network reliability and performance issues are both of great concern to you, try a practical approach, simply to "double-up." Every computer is equipped with two network adapters, and you have for each workstation two network cable connections. You double your hubs, switches, and routers as appropriate. You may consider this overkill, and it probably is, but you should look at it this way: If it is done while building your network, the additional costs will be modest. Standard CAT5e/6 cable used for offices can by itself support two network and two phone connections, while building in a performance enhancement and a reliability upgrade at the same time!

With doubled network cabling and routers you are now also able to have the routers automatically reroute your traffic through an alternative network path if the preferred route is down due to a failed connection. You can also use the mechanism for load balancing, but at that point you would probably need the help of an experienced network consultant. In any case, if you just think about this early in the process, such as when setting up a new office infrastructure, the additional cost to your overall bill will be minimum.

Of course, you should always have a carefully chosen inventory of spare networking components available, so that you will be able to quickly replace a faulty part. And if a network does go down, and you still have not outsourced your systems that need near 24/7 availability, you should install a network monitor that would automatically dial the numbers of the people that are assigned to handle the emergency. By the way, most of these units also monitor environmental conditions, such as loss of electrical power or room temperature as an indication of proper air conditioning. Some can even detect smoke or water in the room and they can therefore be of great help in avoiding subsequent damage.

You should purchase stand-alone network components. It is generally not worth shortcutting expenses by not purchasing additional network equipment and instead configuring a computer as a router or print server, for example. You are building network functionality on a complex piece of equipment that is much more likely to fail than a comparatively simple stand-alone solution.

 

From Prepare for the Worst, Plan for the Best: Disaster Preparedness and Recovery for Small Businesses by Donna R. Childs. Copyright 2008 John Wiley & Sons, Inc. All Rights Reserved. Used by arrangement with John Wiley & Sons, Inc.