SLO¶
TrueWatch SLO is used to continuously evaluate whether a service meets the agreed reliability target. You can select the most recent 7 days or 30 days as the evaluation period, calculate the compliance rate based on monitor anomaly duration or business success count, and notify relevant personnel through an alert policy when the target is not met.
Concepts¶
| Term | Description |
|---|---|
| SLI (Service Level Indicator) | An indicator that measures service quality, such as uptime ratio and request success rate. |
| SLO (Service Level Objective) | The objective that an SLI is expected to achieve within a specified evaluation period, such as "a 99.9% request success rate over the most recent 30 days". |
| SLA (Service Level Agreement) | A service quality commitment agreed upon between the service provider and the customer. The SLO compliance rate can be used to continuously observe service quality. |
| Error Budget | The failure allowance permitted by the objective. For monitor-type SLOs, it is expressed as the tolerable anomaly duration. |
Select an SLI Type¶
| SLI Type | Use Cases | Compliance Rate Calculation |
|---|---|---|
| Monitor | You already determine whether a service is abnormal through monitors and want to measure the proportion of time the service runs normally. | (Evaluation period duration − Anomaly duration) ÷ Evaluation period duration × 100%. |
| Metric | You have collected business counts such as request counts and task counts and want to measure the proportion of successful requests or successful tasks. | Successful count ÷ Total count × 100%. |
For example, for the same payment service, selecting Monitor evaluates how long the service runs normally; selecting Metric evaluates how many of the included payment requests succeeded. Choose the statistical scope based on your business commitment.
Monitor Type¶
Select one or more monitors as the SLI. The system calculates the anomaly coverage duration based on the anomaly events generated by these monitors, with the anomaly deduction expressed in minutes. Overlapping anomaly coverage periods are merged, and overlaps are counted only once; when grouping dimensions are configured, each dimension combination is calculated separately.
For example, if Monitor A is in anomaly state from 10:00–10:05 and Monitor B from 10:03–10:08, and both belong to the same evaluation object, the merged anomaly covers 10:00–10:08, and the deduction for this occurrence is 8 minutes.
Within the selected evaluation period, the following values are calculated after accumulating anomaly deductions:
- Compliance rate = (Total minutes in the period − Accumulated anomaly minutes) ÷ Total minutes in the period × 100%.
- Initial error budget = Total minutes in the period × (1 − Target ÷ 100).
- Remaining error budget = Initial error budget − Accumulated anomaly minutes.
Using 7 days and a target of 99.9% as an example, the period totals 7 × 24 × 60 = 10,080 minutes, and the initial error budget is 10,080 × 0.1% = 10.08 minutes. If only the 8-minute anomaly above occurs within these 7 days, the compliance rate is approximately 99.9206%, and the remaining error budget is 2.08 minutes.
When creating, set the event threshold according to Targets and Minimum Targets; configure Grouping Dimensions when you need to evaluate by project, environment, or service separately.
Metric Type¶
Configure the Total and Successful queries separately to count the quantities of the same business scope within the evaluation period, and then calculate the success rate. For detailed configuration and examples, see Create SLO.
Each detection generates a complete statistical result for the evaluation period, and the page displays the latest valid result and the data cutoff time. The metric type is used to measure the proportion of successful business operations, while the error budget in the form of failure time and duration is used for the monitor type.
Evaluation Period and Detection Frequency¶
Evaluation Period determines the time covered by each round of evaluation and supports the most recent 7 days or 30 days; Detection Frequency determines the interval for updating results and supports 5 minutes or 10 minutes.
For example, an evaluation period of 30 days and a detection frequency of 5 minutes means the service performance of the preceding 30 days is re-evaluated every 5 minutes.