One of the most important and basic features of any performance monitoring solution is the ability to track metric trends over time (i.e., charting of time-series data). Dynatrace uses sophisticated charting to visualize changes in metric trends over time, making it easy to see trends at a glance.
Comparison of trends within separate time periods has long been a challenge, however. When two time frames occur close together in time, you may be able to make such comparisons by looking at a single chart. Often, however, time frames are too far apart in time to compare trends in a single chart. This is where the new Dynatrace Compare view comes into play.
To compare trends across two time frames
- Select Transactions & services from the navigation menu.
- Select a service to analyze.
- On the service overview page, click one of the tiles in the Requests section of the page.
- On the Details page, click the More… button. This reveals several additional types of analysis that you can run.
- Click the Compare button to open Compare view.
Compare view enables you to compare critical service-request metrics (Response time, Failures, CPU, and Load) across two time frames. Compare view includes a header that contains filters and a comparison time-frame picker that you can use to select a time frame for comparison (options include 2 hours before, day before, week before, and custom time frame).
The charts in compare view are interactive. Click into a chart and drag to adjust the comparison time frame. You can also zoom in/out of the base chart on the left and/or select specific performance percentiles for comparison (Median, Slowest 10%, and Slowest 5%).
The Compare view example below compares performance metrics across two time frames. The base chart on the left shows up-to-the minute performance measurements. The compare chart on the right shows the performance of the same service hours earlier on the same day, when a serious performance problem was encountered. Both Response time and Failures all spiked during the compare time frame.
A Show details button appears beneath each chart. Click the available tabs to view performance metrics for Requests, Attributes, or Instances. The example below shows us the cause of the detected performance problem in the compare time frame: The median response time of the AdsForBlog request was, on average, 9.31s slower than it was during the base time frame.
The Failures chart also reveals that the CalculateRecommendations request experienced an increased failure rate during the same time frame.
In this particular example, two specific requests experienced problems during the compare time frame. It may happen though that specific service instances may encounter problems. You can view such performance detail by clicking the Instances tab.
To view more details, click a request or instance to filter the view. For example, clicking the AdsForBlog request shows us that, while the overall load on the service dropped, the load for the AdsForBlog request actually increased. We can also see that all requests failed during the slowdown. This may very well be the root cause of the slow down.
We will enhance service compare view in an upcoming release. So, please stay tuned for updates on this feature.