Skip to content

app/vlinsert: optimize OpenTelemetry log processingΒ #689

@vadimalekseev

Description

@vadimalekseev

Is your feature request related to a problem? Please describe

Processing logs in OpenTelemetry format currently has poor performance (around 700 KB/s per core) and allocates on average 3.5 objects per log line.

Describe the solution you'd like

We need to optimize the parsing. General optimization recommendations:

  • Avoid unnecessary pointers or consider an arena-style allocator (see example here). Since the OpenTelemetry format is relatively flat, avoiding extra pointers is the simpler option.
  • Eliminate memory copies during unmarshal by referencing the original buffer. This will require refactoring the code to use []byte instead of string.

These two optimizations would significantly reduce allocations and improve performance.

Describe alternatives you've considered

No response

Additional information

No response

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions