Apparent Network's appCritical 1.0

This real-time monitoring tool automates troubleshooting and provides detailed analysis to help you diagnose intermittent network problems.

May 3, 2006

4 Min Read
Network Computing logo

Intermittent network problems can take a big toll on real-time applications, such as VoIP and video transmissions. An occasional packet reorder or drop is more likely to be noticed with these applications than with a file transfer, but it's hard to diagnose these sporadic incidents. Catching the problem in the act, and knowing when it started and stopped, its type and specific data path can go a long way toward isolating its cause.

Apparent Network's new software, appCritical 1.0, catches such hiccups by probing a path, ensuring a continuous low level of testing that kicks into high gear when a problem occurs. When an error is detected, appCritical increases the frequency of its probes, gathering more information about the current network status to improve diagnostics. Companies can define and set thresholds for packet loss, delay, jitter, bandwidth and connectivity loss against which appCritical can measure performance.

App Critical Architecture Click to enlarge in another window

This kind of troubleshooting automation is especially helpful when the call, or video, or other transfer is taking place on a network that is managed by a service provider. Performance-monitoring products from CA, Empirix and ProactiveNet tend to gather performance metrics from as many devices and applications as possible along a path, and apply time-based correlations to the separate metrics to determine a problem's root cause. This approach works well in many situations, but unlike appCritical's method, requires agents or access to OSs to gather data--something no network service provider will allow.

Other competitors include automated traffic-measurement products from BMC Software, Gomez, Hewlett-Packard, and Keynote Systems, which also run periodic probes, but lack appCritical's ability to turn up the frequency of tests automatically when a problem is detected. End-user monitoring products, such as those from Compuware, HP and Lucent Technologies, passively gather users' experience of a path, capturing path-error impact on user connectivity. They fail to gather data when the path is not being used, however, and cannot work across domains.

AppCritical Setup

The architecture of the product is typical and includes a Java management GUI, Sequencer and Network Intelligence System. The NIS is a server that performs the application logic, data collection, aggregation and crunching and houses the database. The Sequencer resides on any number of networked servers or desktops and sends ICMP packets to an IP host or target, receives the responses, and sends the results to the NIS for processing.Targets are tested differently, depending on what they are. Client, data server, voice server, and handset device types are all supported. Different codecs (G.711, G.723, G.729 or GSM) are emulated using ICMP. Three predefined profiles or SQDs (Service Quality Definitions)--a LAN, a VoiceQuality, and Connectivity definition--help to configure thresholds. Each SQD has a template for cloning new SQD values against which tests can be run.

Each group of tests maintains a history of responses; a log of each SQD violation; and an overarching health view of cumulative results. We ran tests against a number of devices across our labs in both Syracuse, N.Y., and Green Bay, Wis. AppCritical performed as expected, increasing the frequency of its testing in an attempt to capture failures. In our tests of servers and VoIP handsets over IPsec VPN tunnels, appCritical caught transient errors and sorted them by severity, displaying events linked to historical baselines and specific tests.

As with all fault-based network management, the risk is that the number of violations reported can create so much "noise" that the truly bad ones go unnoticed. Although appCritical includes a severity filter to limit fault notifications, it's less powerful than those provided by most competitors, which can include the suppression of duplicate and downstream events. However, appCritical monitors a path, not the devices along it, which reduces the number of possible faults recorded.

Although it seems a bargain compared to some competitors, whose costs can run into the high six figures, appCritical isn't cheap. Expect to pay about $100,000 for an average install. But it's a good value, especially for capturing intermittent failures across networks you don't own.

Bruce Boardman is executive editor of Network Computing. Write to him at [email protected]. 0

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights