In-Depth

Agent-less Appliance Identifies Application Performance Problems

NLayers' Insight creates application behavioral models to help organizations optimize and cut costs

• By Stephen Swoyer
• 07/13/2004

For that reason, start-up nLayers Inc. touts a new spin on a familiar theme: Enterprise-wide application performance management, but—here’s the twist—enabled by means of a discrete hardware appliance, called Insight.

nLayers president and CEO Gili Raanan says that application availability and performance issues are a factor in almost all enterprise environments—and typically consume a substantial portion of the overall IT budget. “Most of their [IT's] budgets in 2004 or 2005 are dedicated to maintaining existing applications while only the least of the budget is spent on new investments,” he argues.

Raanan concedes that the application performance management marketplace is by no means bereft of competition, but says that—in addition to nLayers’ appliance-based approach—the company differentiates itself on other grounds. “In the system we have—an agent-less, passive appliance—it means that we do not run any software, any agent, on any of the production servers,” he comments. “We simply do not run a single packet to your production environment. We do not impact, we do not change, the measured environment.”

Instead, he says, Insight plugs into a network. "We connect to multiple points of your network to standard interfaces that are offered by the switch vendors like Cisco and Junipers,” and monitor application traffic.

“What we do is deep packet inspection, we go through the entire six layers through to seven [of the OSI framework], and we go deep into the payload and not just the head of the packet,” he explains. “We look into the payload and header of the packet looking for a unique identifier of applications, which are essentially the unique fingerprints of an application on a network. When we search for these unique names, we look at a packet or a session and tell you, oh, that’s a BEA Weblogic application server talking to a DB2 database.”

Raanan says that the nLayers appliance can even wring useful information out of encrypted internetwork traffic, in the unlikely event that any exists, that is. “Today, we do not open the encryption, and in future versions we might do that by adding an SSL card or adding SSL acceleration to the appliance. But today, even if the communication is encrypted, usually we are hooked through the network at the core of the data center. So even if it is encrypted with SSL, it still has a lot of information in the header itself, like IP address, protocol name, destination address, etc.,” he points out.

How, exactly, does nLayers propose to address application availability and performance issues? Basically, says Raanan, Insight collects historical information about application usage and creates a time-based three-dimensional model—an aggregate behavior model of all enterprise applications. “Based on our model, we can tell you what’s running on a day, versus what’s running on a night, and once we have that behavioral model of your application, it’s time-based and it can hold a baseline of days or weeks of your data center,” he says.

Insight features a tool called Time Slider that lets customers view changes in a three-dimensional model over time. “When you glide through time, what you would see is that boxes [representing applications] appear or disappear, arrows [representing connections between applications] get thicker or thinner, so the time model is dynamic living and breathing model that represents the behavior of your application environments throughout time.”

Insight also boasts custom analytics that are geared toward several common scenarios, such as server room relocations, mergers and acquisitions, or disaster recovery efforts, Raanan says. In addition, these analytics are augmented by wizard-based best practices: “What we do is we add a bunch of best practices through our wizard-based technology." If a customer is consolidating servers, the wizard may suggest that IT “look at the singular function servers in your network so that you can consolidate.”

Although the Insight launch also doubles as nLayers' coming-out party, the San Jose, Calif.-based vendor isn’t just any scrappy up-start. Raanan spent several years with Israeli Intelligence where he developed Insight-like technology when he was charged with administering its data center. As of launch time, the company claims three paying customers, including software vendor Interwoven Inc. nLayers has also notched relationships with systems and application management powerhouses BMC Software Corp., IBM Corp., and Mercury Interactive Corp.