Forge Bridge functions are very slow (Geographic latency in Europe?)

Dear Community,

Forge Bridge’s invoke('funcName'); seems to add about one second of overhead to the execution time of the backend function itself (as measured in the execution time of the corresponding resolver function). The use case here is a CustomUI IssueContext. The performance difference between tunnelling into a local app and deploying to dev is thereby negligible.

Similarly, initializing Forge GUI-s (of all sorts) also seems to be very slow. Here is an example from a different use case:

Rendering a UI Kit 2 ProjectSettingsPage takes about 2 seconds (AFTER the main Jira app GUI has loaded, so this is the time during which Forge’s gray loading indicator animation is being shown). 1.3 seconds thereof are the execution of /graphql with operation=forge_ui_invokeAuxEffects and effect=initialize. This is a bare-bones ProjectSettingsPage, i.e. I have disabled any code that would fetch data at initialization for the purposes of this measurement.

Is this the expected performance in both cases? Can you recommend me any remedies?

2 Likes

My issue might have to do with geographic latency as I am located in Europe (see here: Forge invoke is very slow)

Since this linked post is almost a year old, are there any updates to this? The latencies I am currently experiencing are basically a blocker for the entire European market.

No customer will use an app in which every single operation takes more than a second. (Btw. REST calls to Atlassian products from the backend are also very slow. Getting a single issue from Jira on an instance with < 10 custom fields configured, with renderedFields, takes about 700 ms)

P.s. I have used the new native Node.js runtime for all measurements.

2 Likes