Networking Blog

Remember the old days when networks typically were in one location, used plain old servers and routers, and were homogeneous in nature? In those days, a network monitoring system could take a static approach, where a member of some well-staffed IT department could manually check out each server and switch one at a time.

But nowadays, there is a lot more IT equipment to contend with, the majority of which isn’t even located within the data center. For example, a retailer such as Pacific Sunwear may have 200 devices within its primary data center, but it has more than 7,000 other devices spread across its nearly 950 stores, including such items as video surveillance monitors, routers, and wireless access points. Meanwhile, an airline manufacturer might have hundreds of pieces of equipment on the shop floor, many of which have their own IP addresses.

And IT departments, particularly at small to midsized enterprises, are more overworked than ever before, so old-school network monitoring tools don’t cut it anymore. Given that finding the right network monitoring solution is yet another worry to pile on top of all the other things a data center manager has to remember, such as data security, mobile device management, and email archiving for compliance purposes (and these three were picked out of a hat holding 100 other issues), what criteria should managers keep in mind for choosing the best solution for their businesses?

Concentrate On The Baseline

Traditionally, network monitoring tools tracked the degree of availability within a network and inferred performance by measuring the utilization of server CPUs, memory, and network traffic, says Steve Harriman, vice president of marketing at NetQoS.

However, none of these issues matter if an end user working at a branch office 1,000 miles away from the corporate network is experiencing a 10-second response time when he or she is accustomed to, at most, a two-second one, Harriman says. Moreover, networks are handling many different types of data, all of which have differing characteristics about which traditional monitoring tools simply are clueless.

According to Harriman, the best measure for understanding how to monitor a network is setting baseline metrics in the server infrastructure for “normal” performance. If you have these baselines in place and discover deviations from these metrics, you will have an easier time determining why that eight-second latency has occurred for the aforementioned end user. “It could be because an application was poorly coded or because the application is chatty, with lots of calls to the corporate database before it gets to the end user’s screen,” Harriman says.

Determining this baseline should not be too challenging, as long as you do not overthink it. For example, voice data has strict requirements that are at the same time easy to measure. “Everyone knows when a call is good or not. If they experience a delay, stuttering, or no dial tone, they are going to hang up,” Harriman says.

Automate

For her part, Yankee Group analyst Vanessa Alvarez says that automation is a key criterion for choosing a network monitoring solution because it increases its functionality while lowering the IT resources required to keep abreast of what is taking place on the network. Automation capabilities also make the solution easier to manage and to integrate with systems as a whole, says Alvarez.

CITTIO’s Lerner says that automation is fundamental in any sophisticated IT environment that has multiple locations, incorporates a range of different devices, and is increasingly heterogeneous in nature. “You don’t really see 100% Microsoft or Cisco [setups] anymore, and a static approach makes things that much harder,” he says.