starred/headscale
1
0
Fork 0
mirror of https://github.com/juanfont/headscale.git synced 2026-01-12 04:10:32 +01:00
Code Issues 98 Packages Projects Releases 10 Wiki Activity

[Feature] Replacing encoding/json with https://github.com/bytedance/sonic #992

New Issue
Closed
opened 2025-12-29 02:27:08 +01:00 by adam · 3 comments
adam commented 2025-12-29 02:27:08 +01:00
Owner
Copy Link

Originally created by @JanisCasper on GitHub (Mar 31, 2025).

Use case

Performance enhancement

Description

Hi,
we've been evaluating Headscale for our company for a while and started to look for quick wins to improve the performance of the server. One bottleneck we stumbled upon is the significant amount of CPU time spent on JSON parsing operations. Although part of the standard library of Go, encoding/json is not the best-performing JSON parser (see benchmarks like https://github.com/go-json-experiment/jsonbench ).
Would it be worth discussing a switch to Sonic? We've conducted some internal load tests, and the results are quite promising.
I’m happy to create a matching PR but wanted to reach out to you first to discuss the matter.

Contribution

  • I can write the design doc for this feature
  • I can contribute this feature

How can it be implemented?

No response

Originally created by @JanisCasper on GitHub (Mar 31, 2025). ### Use case Performance enhancement ### Description Hi, we've been evaluating Headscale for our company for a while and started to look for quick wins to improve the performance of the server. One bottleneck we stumbled upon is the significant amount of CPU time spent on JSON parsing operations. Although part of the standard library of Go, encoding/json is not the best-performing JSON parser (see benchmarks like https://github.com/go-json-experiment/jsonbench ). Would it be worth discussing a switch to Sonic? We've conducted some internal load tests, and the results are quite promising. I’m happy to create a matching PR but wanted to reach out to you first to discuss the matter. ### Contribution - [ ] I can write the design doc for this feature - [ ] I can contribute this feature ### How can it be implemented? _No response_
adam added the enhancement label 2025-12-29 02:27:08 +01:00
adam closed this issue 2025-12-29 02:27:10 +01:00
adam commented 2025-12-29 02:27:11 +01:00
Author
Owner
Copy Link

@kradalby commented on GitHub (Mar 31, 2025):

One bottleneck we stumbled upon is the significant amount of CPU time spent on JSON parsing operations.

We are aware of this bottleneck, it is because we use the database for storage when the application is running, even if all the data could be kept in memory. It is somewhere on the roadmap to rearchitecture this, but we have a lot to do before.

Although part of the standard library of Go, encoding/json is not the best-performing JSON parser (see benchmarks like go-json-experiment/jsonbench ).

The value of the stdlib is stability over time, headscale isnt really a performance project, as often stated, its targeted at home labbers.

Would it be worth discussing a switch to Sonic?

I dont think it is something we would consider. Sonic, as an example, is not pure Go, contains ARM and C code which we cannot look at, evaluate and which would dramatically complicate our builds.

I think these sort of large unknown "short term" wins would not really be beneficial and rather promote us to not address the actual issue.

@kradalby commented on GitHub (Mar 31, 2025): > One bottleneck we stumbled upon is the significant amount of CPU time spent on JSON parsing operations. We are aware of this bottleneck, it is because we use the database for storage when the application is running, even if all the data could be kept in memory. It is _somewhere_ on the roadmap to rearchitecture this, but we have a lot to do before. > Although part of the standard library of Go, encoding/json is not the best-performing JSON parser (see benchmarks like [go-json-experiment/jsonbench](https://github.com/go-json-experiment/jsonbench) ). The value of the stdlib is stability over time, headscale isnt really a performance project, as often stated, its targeted at home labbers. > Would it be worth discussing a switch to Sonic? I dont think it is something we would consider. Sonic, as an example, is not pure Go, contains ARM and C code which we cannot look at, evaluate and which would dramatically complicate our builds. I think these sort of large unknown "short term" wins would not really be beneficial and rather promote us to not address the actual issue.
adam commented 2025-12-29 02:27:12 +01:00
Author
Owner
Copy Link

@JanisCasper commented on GitHub (Apr 1, 2025):

Allright, thanks a lot for answering!
Would you guys be open to contributions (from us) to the rearchitecturing?

@JanisCasper commented on GitHub (Apr 1, 2025): Allright, thanks a lot for answering! Would you guys be open to contributions (from us) to the rearchitecturing?
adam commented 2025-12-29 02:27:12 +01:00
Author
Owner
Copy Link

@kradalby commented on GitHub (Apr 14, 2025):

We are open to contributions that help, but help on larger work where things touch each other, maybe other than code review is a bit hard. In general the overhead of planning quickly seem to defeat the purpose. But independent work or small things (like https://github.com/juanfont/headscale/pull/2514 and https://github.com/juanfont/headscale/pull/2509) is easier.

@kradalby commented on GitHub (Apr 14, 2025): We are open to contributions that help, but help on larger work where things touch each other, maybe other than code review is a bit hard. In general the overhead of planning quickly seem to defeat the purpose. But independent work or small things (like https://github.com/juanfont/headscale/pull/2514 and https://github.com/juanfont/headscale/pull/2509) is easier.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
CLI
DERP
DNS
Nix
OIDC
SSH
bug
database
documentation
duplicate
enhancement
faq
good first issue
grants
help wanted
might-come
needs design doc
needs investigation
no-stale-bot
out of scope
performance
policy 📝
pull-request
Mirrored from GitHub Pull Request
 question
regression
routes
stale
tags
tailscale-feature-gap
well described ❤️
wontfix
No Label enhancement
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
adam
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/headscale#992
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?