But finding the causes of performance problems usually goes beyond the scope of application analysis. Although it provides high-level visibility, the best way to identify the source of performance problems is to view the packets that make up the conversation.
By drilling down to a step-by-step diagram of the conversation between client and server, you can often identify the precise point in the transaction that is problematic. This detailed conversation view, called Connection Dynamics, is unique to Observer and provides easy identification of long latency and response times and pinpoints network impediments.
Connection Dynamics is especially useful in applications such as VoIP that depend on more than one TCP connection. For example, if VoIP analysis indicates call setup times are increasing, Connection Dynamics will show whether the client, call manager, or network is causing the problem. The diagram below shows the call setup and connections between the call manager and the phone initiating the call.
In a real-world example of troubleshooting applications, Jeff Larkham, senior communications specialist for Weather Services International (WSI), uses Connection Dynamics to manage Internet traffic. WSI uses the Internet as a means to provide clients, such as CNN, United Airlines, and UPS, up-to-the-minute weather information. Without access to weather information, lives and businesses could be in peril.
Connection Dynamics allows Larkham to watch traffic passing through ports and assess the corresponding latency. In one case, he used it to determine whether an Internet problem was caused by WSI’s network or the client. Selected clients get access into weather information servers via their VPN. A client, who had been accessing the service, was suddenly denied access–keeping that organization from getting pertinent weather information.
Larkham viewed Connection Dynamics as the client attempted to log into the network and in real time discovered that the IP address trying to log in was not the same as the registered IP address. That proved that the problem was on the client’s end. In fact, the client’s IP address had been changed by the client’s IT administrator without the client knowing. Larkham was then able to help the client re-register and gain access to the network.
Users like Larkham recognize the troubleshooting value detailed views of application conversations provide. “All I have to do is pull up the response times and follow the conversation to see if the problem is on our end or whether the customer just has a slow Internet connection,” said Larkham.
When high-level application analysis isn’t enough, Connection Dynamics provides the detailed view needed to pinpoint the cause of an application slowdown.
|
