Skip to main content

Cache and performance

This page explains the user-facing side of Aether CLI performance.

What users control directly

Aether CLI users do not control the hosted service, but they do control:

  • local cache location
  • cache sharing strategy
  • metadata TTL behavior
  • logging verbosity
  • metrics visibility

Those settings shape the local experience substantially.

Cache directory strategy

Use a dedicated cache directory when:

  • a machine runs many sessions
  • different environments should stay isolated
  • CI jobs should not reuse personal state

Use a shared cache root only when the sharing behavior is intentional and understood.

TTL tradeoffs

Lower TTLs:

  • improve freshness
  • reduce stale directory or lookup behavior
  • increase remote lookups

Higher TTLs:

  • reduce round trips
  • improve perceived responsiveness in stable workflows
  • can hide remote changes longer

Logging tradeoffs

Higher log verbosity helps with diagnosis but increases noise and may complicate routine workflows.

Use debug logs for:

  • mount failures
  • auth confusion
  • unexpected local behavior
  • support escalations

Metrics tradeoffs

Metrics are most useful when:

  • performance is inconsistent
  • a mount feels stale or slow
  • support needs a local view of the client

If you do not need them, keep the metrics endpoint simple and local.

Practical performance guidance

Use this sequence before changing many knobs:

  1. confirm endpoint and auth are correct
  2. isolate cache roots by environment
  3. lower TTLs only if freshness is the actual problem
  4. enable metrics or debug logs only when diagnosis is needed

That order avoids “tuning” the wrong problem.