I want to know if you can help me. we want to monitor the performance of MWG, and improve the user experience.
We usually see statistics like the following:
What does exactly corresponds the statistics of Detailed HTTP Performance, we see that there is no correlation with the General Performance, Why? What are normal values for this statistics? and what we have to look if we see peaks of processing time?. These are some questions that we like to answer.
I think both statistics show different data so you cannot correlate or compare what they say. The "General Performance" gives a very rough overview about how long it takes to make a DNS lookup, how long connecting to a server takes, (e.g. MWG attempting to make a connection to a web server until the connection is established) and how long the request was processed in the rule engine.
The second one shows the amount of time taken for transferring amounts of data. When you see a spike there it means that upload and/or download took some time, probably the server was not very fast or there was a bigger amount of data that was transferred. The time for transferring data is not part of the "General Performance" chart.
Assuming you want to download a 5 MB file:
10 ms DNS Lookup -> General Performance Chart
500 ms Connecting to Web Server -> General Performance Chart
50 ms filtering Request -> General Performance Chart (Time in Rule Engine)
25000 ms downloading the file -> Detailed HTTP Performance
100 ms filtering Responst -> General Performance Chart (Time in Rule Engine)
I cannot say what normal values for these statistics are. This is something probably other users of MWG can share.
For peaks in General Performance:
- If DNS Lookup peaks out make sure DNS is fully functional, fast and not overloaded in any way
- If Connect to Server peaks out there is actually nothing you can do because it means MWG cannot connect to a remote server very quickly. If you know your network is generally operational and fast it is most likely that this spike is caused by the remote end.
- If Used By Rule Engine peaks you should make sure MWG is not overloaded but Load and CPU usage are "OK" and there were no recent changes in the rule engine which cause an increased processing time
For peaks in Detailed HTTP Performance:
Those depend very much on the amount of data transferred. Because the statistic does not tell you the average speed of the transfer but only tells you how long transferring data took you don't know if the 300.000 spike was for a 500 MB ISO image (which would be OK) or for a 5 KB GIF (which would indicate a problem). Therefore I am not sure if the spikes here are very helpful for troubleshooting.
Do you actually encounter any problems with that MWG installation?