Enterprise networks are becoming more complex and are carrying more data than ever, putting more stress on network administrators and the tools they use to monitor network performance. A stoppage or slow-down could have serious material effect on a business, so it is critical to identify problems as quickly as possible or, even better, to anticipate them.
Startup company Sideband Networks has announced a hardware/software product that can monitor and analyze network data traffic in a virtual environment in real time while comparing it to previous network activity patterns. Comprised of the XRE-4000 hardware appliance and the vXRE virtual appliance, Sideband developed its product to allow network admins to react faster to incidents, and it does so in part due to its predictive alerting ability. For example, the company says its appliances can recognize activity that in the past has preceded a problem, and then notify an admin to take corrective action to prevent the problem from occurring.
“Our solution provides a means of optimizing and providing capacity planning for a critical-to-use resource,” says Zane Taylor, Sideband Networks CEO. Using the network topology discovery and analytics, he adds, the Sideband products provide a means of audit and assurance. With detailed analytical correlation and the means to alert and take action, the IT administrator can proactively ensure network capacity and performance. About 10 companies are currently using the Sideband product in alpha trials according to Taylor.
The challenge to using predictive analytics to anticipate network problems or even identifying the cause of a problem has been the structure of a typical enterprise network—a connected collection of multiple systems spread over multiple business groups and geographies with virtualized and cloud-based elements. “The way networks have been built, it’s hard to tell what caused a problem,” says Lee Doyle, principal analyst at Doyle Research. “You put out probes and get all the data you can, but it’s hard [to know what that data means], especially at 10 gigabytes per second.”
Sideband’s answer is to create a map of the enterprise network using what it calls “dynamic network discovery.” The resulting diagram shows all physical and logical connections and provides a visual for showing anomalies or problems. Rather than analyze all traffic data, it samples the data using heuristic filtering, allowing the analysis to occur in real time at network speeds up to 40 gigabytes per second. A network admin can see the results through a dashboard like the one shown below.
Operating expenses represent 75 to 80 percent of the cost of running an enterprise network, says Doyle. “Most organizations rely heavily on their networks, not just from the customer side but for internal production as well. Lower latency and higher reliability will definitely help [reduce operating expenses].”
Doyle sees this ability to combine historical and live data as a key benefit to network administrators. “[Admins] will be able to easily set policies and get actionable information rather than hunting for a needle in a haystack. Network managers are now not being reactive. They now have the opportunity to be more proactive and go after more interesting challenges.”
Michael Nadeau is the publisher of Data Informed. Follow him on Twitter: @menadeau.