Slower Jira fetch responses for deployed Forge app vs forge tunnel

I was conducting some performance testing for a Jira Cloud Forge app and ran into a strange issue: running the app via forge tunnel results in better performance than when it’s deployed and running directly from Forge.
Non-HTTP operations took longer (explainable by throttled resources) but also Jira API fetches take longer when initiated from the deployed Forge app. This makes no sense to me, but I confirmed that our requests to Jira (ex. api.asUser().requestJira(jiraRoute, requestInit)) take twice as long to return when deployed vs. tunnel.

Has anyone else encountered this?
Is there a logical explanation?

Thanks!

1 Like