Skip to content
Start free

Know when the error budget burns, and who answers

Set availability and latency objectives on the traces or logs you already send, and every SLO gets its own burn-rate monitor. On-call schedules and escalation policies carry its pages to the right team until someone acknowledges.

Start freeBook a demoRead the SLO docs
  • 7, 30 or 90 days
  • 14.4× · 6× burn alerts
  • 5-step escalation

How it works

From an objective to an acknowledged page

The SLO, its burn-rate monitor and the escalation behind it form one chain, from the first minute of burn to the person who acknowledges the page.

  1. 01DefinePick traces or logs, availability or latency, a service, a target and a 7, 30 or 90-day window.
  2. 02MeasureGood and total events roll up every minute; each SLO shows its budget and its burn over 1 and 6 hours.
  3. 03AlertIts burn-rate monitor goes critical at 14.4× and warns at 6×, over the last hour and 5 minutes.
  4. 04PageA notification policy with an escalation policy opens a page for the on-call schedule's team.
  5. 05EscalateAn unacknowledged page moves to the next step after the wait you set, up to five steps.
  6. 06ResolveAcknowledging stops escalation, and the page resolves on its own when the alert recovers.

SLOs

Objectives on the data you already send

An SLO counts good and total events: spans without errors or under a latency threshold, or log lines below error. It rolls them up every minute, and the error budget and burn rates follow from the target you set.

  • Scope an SLO to one service, environment or service group, or leave it open
  • Minutes with no events are left out of the SLI instead of counting as good
  • A deploy fails its post-release check when its service burns an SLO faster than 6× over an hour

Escalation

Pages climb the policy until someone acknowledges

Each escalation step names the schedules, teams or channels to page, and how long to wait for an acknowledgement before it moves on. Acknowledge the page and the climb stops.

  • Rotations

    Members hand off every 1 to 90 days, weekly unless you choose otherwise.

  • Overrides

    Take over a shift in one click; the latest override always wins.

  • Up to five steps

    Each step waits from 1 minute to 24 hours, 15 minutes by default.

  • Named in every page

    Pages go to the team's channels and say who is on call.

On-call

Who is on call now, and what paged them

The On-call page shows every schedule with the person on call and when their shift ends, the escalation policies in use, and the pages of the last seven days with who acknowledged each one.

  • Take over the current shift in one click when you swap with a colleague
  • Every page keeps its step, its acknowledger and when it resolved
  • Ask Brain who is on call at any moment, or to acknowledge a page for you

Better together

Reliability goals tied to the signals behind them

SLOs read the same spans and logs as the rest of your workspace, so a burning budget leads straight to the traces and the problem behind it.

FAQ

Questions on-call teams ask first

Talk to an engineer
Does Tracix call or text the person on call?
Not yet. A page goes to the channels of the team behind each schedule, such as Slack, Microsoft Teams, PagerDuty or Opsgenie, and names who is on call. There are no SMS, voice or mobile app notifications today.
Can we import schedules from PagerDuty or Opsgenie?
No. Schedules and escalation policies are Tracix objects you create in Tracix. You can still route pages to PagerDuty or Opsgenie as channels.
How does missing data affect the error budget?
It does not. Minutes with no events are left out of the SLI, and if the whole window is empty the SLO shows no data rather than a perfect score.
Can a schedule have several layers or shorter shifts?
Not yet. A schedule is one rotation that hands off every 1 to 90 days, with overrides for exceptions, and handoffs are calculated in UTC.

Set your first SLO

Pick a service you already trace, choose a target and a window, and its burn-rate monitor is created with it.

Start freeRead the guide