Biggest lie in the enterprise: 'The network is down'

Stop blaming the network for outages -- applications and Microsoft operating systems are at fault over 99% of the time

This article was contributed by a reader. If you have an opinion or technology experience you would like to share, contact Online Community Editor Julie Bort (jbort@nww.com).

Enterprise network professionals take note: If you are considering doubling the size of your flexible-spending medical account to buy more Ibuprofen, antacid and sleep aids, I’ve got news for you. None of these is a cure for the curdled stomach and throbbing head caused by this dreaded message: “The network is down.” Knowing that the message is not true 99.9x% of the time is not the cure either.

As the director responsible for the enterprise network of a midsized healthcare organization, I can sympathize. Our telecommunication technology department services a converged network for data, voice, video and physical security access control. The network spans 25 buildings in a 40-mile circumference linked over private fiber. During the hurricane seasons of 2005 and 2006, our small town took three direct hits. But we experienced no loss of network services.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

This article was contributed by a reader. If you have an opinion or technology experience you would like to share, contact Online Community Editor Julie Bort (jbort@nww.com).

Enterprise network professionals take note: If you are considering doubling the size of your flexible-spending medical account to buy more Ibuprofen, antacid and sleep aids, I’ve got news for you. None of these is a cure for the curdled stomach and throbbing head caused by this dreaded message: “The network is down.” Knowing that the message is not true 99.9x% of the time is not the cure either.

As the director responsible for the enterprise network of a midsized healthcare organization, I can sympathize. Our telecommunication technology department services a converged network for data, voice, video and physical security access control. The network spans 25 buildings in a 40-mile circumference linked over private fiber. During the hurricane seasons of 2005 and 2006, our small town took three direct hits. But we experienced no loss of network services.

Our network is completing the fourth major upgrade in 15 years. Yes, we have more sites and more users, but that’s not the only reason driving us to move from 10M to 10G. We also are upgrading because inefficient applications have helped create a fivefold increase in chatty broadcast traffic, which now comprises 25% of all network traffic. While network devices have become more efficient at processing packets, applications and operating systems have not.

Now, I ask: Should network departments continue to increase bandwidth or take on network application optimization to compensate for poorly written code? I believe the best solution is for IT departments to demand that developers of operating systems, applications, hosts and end devices provide more efficient products.

With the move to Intel-based server/hosts and personal computers, somehow the network got stuck with the expectation of compensating for Microsoft’s operating systems. The MS OS is used in the mainstream while bugs continue to be found and patched. Before the fix process is complete, a new OS release will force the process to begin again. Is there any enterprise network that has not been down because of a DOS attack propagated by poorly patched PCs or Servers? A manufacturer of network equipment would not survive with a level of bug patching equal to Microsoft’s.

But when a customer cannot make a phone call, connect to a server, access a database, send a document to a printer or browse a Web site, in the user’s mind, “the network is down.” The true cause more often than not had nothing to do network failure. A VoIP phone may have lost a file and cannot authenticate, a specific server on the intranet could have been down, or the Web site that was browsed was not available. No matter. As users see it, that “the network is down.”

We, the keepers of the network, know for a fact that network availability is consistently 99.9X%, meaning that only a few thousandths of a percent of the time is the physical network at fault when a user experiences trouble. Gigabit Ethernet switching runs on redundant equipment and redundant fiber; network-management systems monitor traffic in real time; network analyzers, network-access control and intelligent-network event logs continue to enhance reliable network service.