HP Analysis:-
The goal of analysis is to find out bottlenecks in the
application under test and then investigate the source of bottlenecks.
1)What was the average transaction response time?
2)What was the status of server resources when response time
was high?
3)Did we meet the SLA?
4)What component of the system is responsible for decrease in
performance?
5)Can we investigate the possible cause of the bottleneck by
correlating the performance related data of the system and providing the
solution to overcome it?
“After analyzing the Average Transaction Response Time
graph, we investigate the reason for high response time at particular instant
and using graph auto correlation we come to conclusion that there was some
system memory shortage at that particular moment”.
HP Analysis Graphs
Analysis tool shows reports and graphs of the data captured
from system under test. There are many graphs like,
Running Vusers
Hits per Second
Throughput
Average Transaction Response Time
Windows Resources
Let us focus on Average Transaction Response
Time graph,
By observing the graph you can say that, the average response time of check_itinerary transaction fluctuate with big difference. It goes to peak value at certain time. If everything is working fine then average response time should be relatively stable.
Use Filters to see what exactly you want to see in graph.
By observing the graph you can say that, the average response time of check_itinerary transaction fluctuate with big difference. It goes to peak value at certain time. If everything is working fine then average response time should be relatively stable.
Use Filters to see what exactly you want to see in graph.
Merging Graphs
You can merge graphs to see the effect of one parameter on
other. There are 3 ways to merge graph,
Overlay: View contents of 2 charts that share a common X Axis.
Tile: View the contents of 2 charts that share a common X
Axis, one above the other.
Correlate: Plot Y axis of two charts against each
other
You can see in above figure that we have merged
the Average Transaction Response Time and Number of
Vusers using correlate merge. The average response time increases as the
Vusers increases.
When user reaches to 63, 64 there is a sudden raise in the response time, from this you can say that there must be some issue with application so that response time got raised.
When user reaches to 63, 64 there is a sudden raise in the response time, from this you can say that there must be some issue with application so that response time got raised.
Graph Auto Correlation
Click on the Average Transaction Response
Time graph. Look at the check_itinerary transaction, you can see
that average response time started to increase at time slice of 2-4 minutes.
To auto correlate, right click on the check_itinerary transaction and choose Auto Correlate…
Auto Correlate pop-up will show up. Set the time range according to the peak value range of the response time.
In correlation options there are many components from that select Windows Resource.
When you are done with all settings click on Ok to auto correlate,
From the graph you can see that, Private Bytes and Pool NonPaged Bytes are very high at same time slice. So by this you can say that there may be shortage of system memory at that particular moment when response time forcheck_itinerary transaction was high.
To auto correlate, right click on the check_itinerary transaction and choose Auto Correlate…
Auto Correlate pop-up will show up. Set the time range according to the peak value range of the response time.
In correlation options there are many components from that select Windows Resource.
When you are done with all settings click on Ok to auto correlate,
From the graph you can see that, Private Bytes and Pool NonPaged Bytes are very high at same time slice. So by this you can say that there may be shortage of system memory at that particular moment when response time forcheck_itinerary transaction was high.
No comments:
Post a Comment