When making response time calculations, the management console assumes that it is positioned next to the application server. Monitoring response time close to the server enables the management console to accurately measure network and server response time.
Monitoring a web site on the internet will result in skewed metrics because the management console is next to the client, not the server. Server Response Time measured at the client will include the network latency to the server.
When monitoring an internet-facing web application, we recommend that you disable pWhen monitoring an internet-facing web application, we recommend that you disable performance thresholds on network performance metrics and only monitor performance threshold status for server performance metrics. Thresholding for network metrics makes sense for client subnets which are specific to a particular site and which share a particular circuit. The internet is too broad a category for meaningful network performance thresholds.
If you want to monitor a third-party service like google.com, use an IPSLA test to record response time; however, there is no way to know when a spike is due to Internet latency, server response, or an application issue.
|
Copyright © 2014 CA Technologies.
All rights reserved.
|
|