HTTP Reverse Proxy Caching - what to cache?

Hi folks,

this is a hybrid question between infrastructure-related and application-related topics. I’ve approached Atlassian’s support recently because I wonder which types of content can be cached from an Atlassian application (be it Jira, Confluence, etc.). Next to the FR for some official documentation ([JRASERVER-67353] Recommendation on HTTP caching - Create and track feature requests for Atlassian products. - feel free to also vote if that’s something you’d also want to have) they also recommended to get in touch with the Developer Community.
The problem is that, as far as I’ve had to learn, there is a lot of stuff you can cache flawlessly (e.g. images or some stylesheets) but also a few things that seem to be static but are dynamic (e.g. batches that are regenerated upon addon installation). Can you guys help me out with sharing your experiences and what I could consider to create a high-performance, but not too error-prone caching setup?

Warm regards
Peter

1 Like

Craig Castle-Mead has thankfully recently summarized his related exploration and promising results in CDN/CloudFront and Atlassian Server apps.

I have yet to implement this for our own production instances due to a complication via an additional auth layer, but his configuration seems to be a good starting point that can probably be further refined over time (for the AWS context, we might try to codify and evolve this configuration as a CloudFormation template, though ideally it should be a feature of Atlassian’s own Data Center quick starts in atlassian-aws-deployment of course).

Thanks, Steffen. That’s sort of the thing we were looking for. We’ll also evaluate the benefit of these settings but indeed, they look promising.