When performing many modifications to your AEM Java Content Repository (JCR) does your instance slow down, perhaps to the point where it is unusable? What is causing the slow down? Can it be prevented?
The JCR is usually tuned to be quick with read operations but is relatively slow with writes. Many quick, subsequent writes can be quite taxing to the JCR session and the instance. But how often should you perform session saves? Can you save too infrequently?
There are articles on this problem, many giving advice on the perfect save frequency. Is there a one-size-fits-all number?
With these questions in mind, we’re presenting a new tool for testing session save frequencies:
We hope this repository will help show, in a hands-on way, how different frequencies of JCR Session Saves can affect overall performance and instance health. It may give insights into your AEM server.
Our goal is to help drive in the importance of tuning the frequency of JCR Session saves. Just because it “works” in development, does not mean it will be efficient or safe in production as the content grows and grows. A few minutes of consideration during development can save an instance’s health and save time attempting to solve the problem after deploying.
While you are looking at the repository, also check out the other experiments on performance. If you have ideas on other helpful experiments, we would love to hear from you in the issues, or even offer a pull-request.
Thanks for reading. We hope you find these AEM Performance Experiments helpful.