Service Level Objectives
TPX maintains service level objectives for the Partner API. These targets are measured continuously on our side and inform capacity planning; they exclude partner-side network latency.
Availability
| Metric | Target |
|---|---|
| Uptime | 99.9% |
| Error rate | < 0.1% of requests |
Planned maintenance windows are communicated in advance and excluded from availability calculations.
Responsiveness
General targets by operation class:
| Class | Target |
|---|---|
| API reads and single-order writes | Sub-second in the typical case — quotes, submission create/amend/cancel, queries |
| Bulk operations | A few seconds, scaling with batch size |
| Strategy lifecycle | A few seconds — activation provisions demand across every product a strategy targets |
| Webhook delivery | Within seconds of the event; failed deliveries retry with backoff — see Webhooks |
| Real-time pushes | Near-instant — fills and order updates publish to the socket as they happen |
Out of scope
- Settlement timing — executions settle asynchronously; settlement is a state machine, not a latency guarantee. See Order Executions.
- Catalog feeds — regenerated on a daily cadence, not latency-bound. See Feeds.
- Fair-value refresh — models update continuously, but refresh frequency varies by asset type and market activity.