Graph Correlation

Micro Focus Analysis Tool has graph merging option. This option merges the graphs and provides a single view which is helpful for issue identification.  

Graph correlation is about establishing a relationship between performance metrics by comparing the data. The software system is always complex with multi-tier architecture, different technologies and interfaces with internal or external systems. So the easy way to do correlation is to compare the end-user performance with the server-side metrics. It is like comparing the trend from one set of metrics like Response time with other metrics like webserver CPU utilization.

What do we get by correlating the graph?

When you see a relationship between two sets of data like an increase in response time corresponds to increase in no. of users and heap size that uncovers an area for further investigation and analysis. Refer to the below graph:

Figure 01: Graph Correlation

It shows an association between client-side metrics and the server data. The graph shows that when no. of users become 40 then a sudden increase in heap size and similar we can see when users become 70 and 100. It shows that the garbage collector needs investigation. 

In a similar fashion, you can merge the client and server-side graphs to identify the exact bottleneck.

What are the best graph combinations in Correlate Method?

Case 1: High Response Time without error

Correlate Response time graph with Data Throughput graph, Memory Utilization graph and DB Query Processing Time graph

Case 2: Error

Correlate Error graph with Response Time graph, CPU Utilization graph, Memory Utilization graph and DB Query Processing Time graph

Along with the above correlation of the graph refer Heap Dump, Thread Dump and GC for more clarity on the root cause.

Note: 

1. In an ideal scenario, if metrics are directly proportional then y-axis graph lines should follow each other. Else the opposite trend should be followed when metrics are inversely proportional.   Example: During steady-state, Throughput and Hits/ sec graph follow each other while Errors/sec and Average Response Time graph follow the opposite direction
2. A linear line in the forward direction indicates good results and the stability of the application whereas step-up, step-down and spike in the correlated graphs lead the investigation.