Troubleshooting production issues with dynatrace
While I am trying the free trial and integrating dynatrace with my application, I am interested in understanding how would dynatrace help me out in pinpointing the root causes of the problems (in addition to sharing metrics) and based on that what steps should I follow to configure dynatrace.
Below are some of the starting pointers I would like to get insights on
- Detect long running query executions over time. We have processing jobs running periodically in the background. Over time some queries take longer to execute due to various factors - data changes, environment changes etc. Can APM monitor this?
- Disk usage detection. Can the APM solution determine what is the root cause when the disk usage is high?
- Exception traces – Can the APM solution pinpoint me to the code base that could be the root cause of some of the bugs that appear only in production environment? Can it record the series of operations/method calls that the user performed before the exception occurred?
- Can it pinpoint to network failures?
- High resource usage events – high cpu usage, memory leaks, frequent garbage collection runs. How does the APM solution help me in troubleshooting these? Does it automatically take thread dumps when such an event occurs and analyze it?
Kindly let me know your thoughts
Hi
Let me answer your questions:
1: Yes. dynaTrace traces every single SQL Execution. You can get alerted in case DB Execution time exceeds a certain threshold and then get down to the actual SQL Query including bind values and how long it took. dynaTrace is also able to show you these execution times in dashboards over a longer period of time. that makes it easy to identify certain access patterns, e.g: always slow on monday mornings. Here is a blog that shows how to analyze DB related performance problems in production: http://apmblog.compuware.com/2014/04/09/sync-your-timeouts-when-load-balancers-cause-database-deadlocks/
2: Yes: dynaTrace automatically monitors Disk, CPU, Network and Memory of each machine where you have at least one agent installed. We also automatically alert in case you are running low on disk space. We also have a nice blog that shows how to analyze these types of issues: http://apmblog.compuware.com/2013/06/04/how-to-accurately-identify-impact-of-system-issues-on-end-user-response-time/
3: Yes: dynaTrace traces every single Exception and records it in the context of the user that executed this transaction. You actually get the full users history of events until that error happened including where the error happened and full stack trace. Check out this blog post for exception tracing in production: http://apmblog.compuware.com/2014/04/01/dont-trust-your-log-files-how-and-why-to-monitor-all-exceptions/
4: Yes: as mentioned in #2 we automatically monitor network interface utilization. On top of that we also identify network issues that impact the application by tracing the resulting Exceptions that are thrown. If you want to analyze ALL network traffic in your system I can also recommend you look at our network performance solution. More info here: http://www.compuware.com/en_us/application-performance-management/products/application-aware-network-monitoring.html
5: Yes: dynaTrace automatically monitors GC and Memory Utilization of all of your processes where you have injected an Agent. In case you detect growing memory or high GC Utilization you can trigger a memory heap analysis finding out which objects are causing memory leaks and which ones cause high GC. dynaTrace also automtacally takes a full Java Heap dump in case your JVM crashes with an Out Of Memory Exeption. Thats great as you have all the evidence on-hand instead of waiting for the next crash to happen where you might take a snapshot just before you run out of memory. Same is true for CPU. We show you which methods contribute to CPU Usage. But not only that. We also show you which transaction/users are calling these methods. And one more blog post on memory diagnostics in production: http://apmblog.compuware.com/2014/02/26/memory-leaks-load-balancing-and-deployment-settings-testing-lessons-learned-from-the-ecommerce-industry/
Hope this helps
Andi
