One of the most feared issues a customer can face in an enterprise application is poor performance, which can be caused by environmental issues, configurations, misuse of the tool, etc. It can be a long journey of tedious tasks—troubleshooting, support tickets, log and evidence collection, steps to reproduce, etc., all in an attempt to discover the root cause.
At CA World ’17, CA PPM’s new out-of-the-box suite of portlets for monitoring and analyzing technical performance issues was demoed.
That’s right! It’s never been so easy to spot the causes of performance issues, thanks to CA PPM’s friendly and dynamic graphical representation.
I can just hear the crowd’s excitement. Read on for more details.
In what CA PPM version are the new log analysis portlets available?
They’re in CA PPM 15.2 and higher versions.
How do they work?
The portlets automate the process of parsing application server log files and storing them in the database for analysis. Portlets are meant to be used to identify performance bottlenecks, frequently accessed features, response times, peak hours, and more.
Portlets are built using NSQL queries against LOG_ tables such as LOG_SUMMARY. This table is a summary of data collected in tables such as LOG_DETAILS and LOG_SESSIONS. Tables are populated when a Tomcat access log import/analyze job is completed.
What preliminary steps should I take?
Rights: Only administrators should be allowed access. Ensure that your administrators have the following rights:
• Administration - Application Setup global access right
• Administration - Access to access the administrator menu
Administrators must be added to the Log Analysis Access group to access the page.
· Ensure that administrator users have been granted rights to run jobs:
- Jobs - Access global access right
- Jobs - Run - All (or at least the appropriate instance level for running Tomcat access log import/analyze job)
How do I run the job?
Run Tomcat access log import/analyze.
· Run on scheduled mode with Log Date blank: The default value will be the previous day.
· Run on immediate mode: enter today’s date as the Log Date value.
Note: To troubleshoot a specific issue, it’s recommended to run the job on demand immediately so that collected data includes the most recently generated line in the logs.
How are the results presented?
Go to the Administration menu in the Security and Diagnostics section and click on Log Analysis.
The page contains four tabs:
- System stats by hour/day
- Average response time by server
- Session counts by server
- Slow actions by day
Sample: Average response time by server
Sample: Slow actions by day
System Stats by Month:
Ø Monthly page views
Ø Monthly page views by server
Ø Monthly session counts
Ø Monthly XOG requests
Ø Monthly XOG requests by server
Sample: Monthly page views
Sample: Monthly session counts
Daily Page/XOG Views:
Ø Slow pages top 20
Ø Page views by service
Ø Page views by server
Ø Average response time by day
Ø Daily page views
Ø XOG requests by service
Ø XOG request by server
Ø Average XOG response rime by day
Ø XOG call by day
Sample: Slow pages top 20
Sample: Average response time by day
Log Data Export: Each row of exported data shows the date, total page views by all users, average page response time in seconds, total XOG requests, average XOG response time, and unique session count for that specific day.
Log analysis is a graphical tool, a reading from the logs. Portlets are self-explanatory. No special guidance is needed to understand them. Just bear in mind that you need to set values in the filter to show data. You must always select a Date, Environment and Hostname from the filter bar.
In my experience, initial root cause analysis of a performance issue takes 5-20 minutes with this tool, compared to several hours reading logs and asking questions. This new component is a true time saver.
For readers interested in more detail, check out DocOps. I encourage you to participate in the best-in-class site, where you have access to your peers, events and support. You can also reach out to CA Services for information about CA PPM Release 15.2 upgrades/implementations, and individualized business outcome references and analysis. Feel free to post in the comments section of this blog or contact me directly via email and Twitter @aurora_ppm.