Monitoring health

THE PROBLEM WITH mission-critical networks is that they meet their mission-critical requirements only while they’re working. Keeping them working, however, means dedicating a staff to that process — after all, these things don’t run without attention.

NOCPulse aims to meet that need through outsourcing. Its product consists of specialized software in the company’s NOC (network operations center) in Sunnyvale, Calif., as well as on a processor, called a Satellite, that resides in your internal network and monitors the health and activity of your network infrastructure. Alerts are sent out when infrastructure activity crosses certain thresholds, and network managers can access the monitoring data online.

Unfortunately, NOCPulse reduces its usefulness by restricting the information you can know about the Satellite and limiting the types of infrastructure you can monitor. NOCPulse is capable of monitoring most infrastructure hardware from Cisco, Foundry, and Alteon and can monitor Windows, Linux, and most flavors of Unix. But it provides only minimal monitoring of other products. Worse, there’s nothing a network manager can do directly to determine whether the NOCPulse Satellite is functioning properly.

Despite these limitations, however, the NOCPulse system works well and does provide the instant alerts it promises.

The NOCPulse Satellite is a secure, 1U, Intel-based computer that runs Linux to retrieve operational metrics from the infrastructure. It’s so secure that there is effectively no means of access for your network management staff — in fact, it won’t even respond to pings.

Although you can access the Satellite through its console port, you’ll need a password, and NOCPulse keeps that a closely guarded secret. Instead, NOCPulse strongly recommends that you keep a second Satellite processor on your network in a fail-over mode. That way, if your original Satellite goes out, the other can take over seamlessly.

During testing, we found the Satellite hardware to be sufficiently unreliable enough to make this fail-over processor a necessity. The company says that an improved version that should address these reliability issues is currently shipping.

The Satellite communicates with the NOC through a special port on your firewall. The metrics it gathers from your infrastructure are passed on to the NOC, where they are analyzed by software and are made available to your staff via the NOCPulse Web site.

If the software detects that critical metrics thresholds have been crossed, it can send out alerts via phone, email, or pager. More serious problems, such as the failure of the Satellite, are handled by the NOCPulse staff. The company staffs the NOC 24 hours a day, seven days a week, and you can also call the staff if you have problems or issues related to NOCPulse.

During testing, we had two significant failures for NOCPulse to pick up. In one case, our ancient Intel Express switch passed its last bit and quietly faded into dead-switch land. Unfortunately, it didn’t turn out its lights, so there was no immediate way to tell it had died. The NOCPulse software knew right away, however, and sent an alert to the Research in Motion BlackBerry that was selected to receive notifications.

In another case, the lab’s broadband connection went out, and for this we received a call from the NOC. Unfortunately, the Satellite was supposed to have reverted to an analog dial-out to report network outage, but that feature wasn’t working at the time of the failure thanks to a “bug” that kept it from dialing. The NOCPulse staff figured out the problem when the Satellite stopped reporting and they couldn’t reach it from their end.

When you get to the company’s Web site to check on the monitoring reports and data, operations information is clear, detailed, and comprehensive, even for those devices for which NOCPulse offers only limited support. Command Center’s software is flexible enough that you can choose the metrics to monitor, the thresholds, the display details, and how you want the reports to appear.

There are two problems with the NOCPulse operation: The first is that changing Satellite and monitoring configurations is not an intuitive procedure. Updating the Satellite is a two-step process, and choosing the right buttons to press isn’t guided by the software. The process of adding or changing monitoring parameters is not particularly transparent nor is it particularly fast.

Second, the NOC may be staffed around the clock, but that doesn’t always translate into usefulness. We tested this product during the first week of July, only to find that NOCPulse had reduced its staff. This meant that the staff consisted of an engineer with a cell phone. The engineer had to first get to a location where he could log in to the NOC before he could tackle the problem. Once he was logged in, things were fine.

Overall, the NOCPulse Command Center is an effective service. You get a level of monitoring that you’d be unlikely to accomplish on your own, probably at a lower cost than would be required for you to do it yourself. In the process, you are spared the problem of finding, hiring, and keeping a staff, and you will get a level of technical sophistication beyond what you’d accomplish yourself.

The problems related to staffing response at NOCPulse need to be resolved, however, as does the inaccessibility of the Satellite. To some extent, those problems will be solved in late August when the company releases the licensed version of the product, which will give companies the option to move the monitoring in-house and thereby have more control over what’s happening.

Either way, there’s nothing else quite like it. If you can live with its limitations, NOCPulse Command Center can play a crucial part in your network management plan.