hashicorp / hashicorp/go-plugin
[Questions] How well does this architecture scale with many plugins?
- Dominant language
- Go
- Stars
- 6.1k
- Forks
- 511
- Avg merge
- 22h 47m
- Merged PRs (30d)
- 4
Description
Great work with this plugin system. Do you have an idea of the performance comparison between running a Go application with no plugins that performs that same amount of work as a Go application with say 500 plugins (all in Go) that all have to be used on each request (either in parallel or serially)? I'm looking for metrics on:
- overhead in terms of RAM/CPU per plugin
- latency startup/shutdown time per plugin
- latency of sending data back and forth between application and plugin (its looks like gRPC is designed for low latency, highly scalable systems)
This is something I could probably test in a few hours, but just curious if there are metrics that you have already available.
Contributor guide
Assessment
This issue has not been assessed yet.