The slo.yaml file contains definitions of the Service-Level Objectives (SLOs) defined for your Keptn installation. The Service-Level Objective (SLO) configuration specifies a target value or range of values for a service level that is measured by Service-Level Indicators (SLI).
Example of Service-Level Objective (SLO):
spec_version: '1.0' filter: mz_id: "4711" svc_id: "a14b-cd87-0d51" comparison: compare_with: "single_result" include_result_with_score: "pass" aggregate_function: avg objectives: - sli: response_time_p95 pass: - criteria: - "<=+10%" - "<1000" warning: - criteria: - "<=800" total_score: pass: "90%" warning: "75%"
The filter property allows a list of key-value pairs that are used
to uniquely identify a deployment of a service.
This means that the key of a filter can be used as a placeholder in an SLI query.
For example, the filter
specifies a unique identifier of the deployment of a service.
Consequently, the key of the filter (i.e.,
svc_id) can be referenced in an SLI query by
The filters project, stage, service, and deployment
can be inferred from the Keptn configuration by using
$DEPLOYMENT, respectively, in SLI queries.
These values can also be overwritten in the configuration. The default filters are:
Example of an SLO with a list of filters:
spec_version: '1.0' filter: mz_id: "4711" svc_id: "a14b-cd87-0d51" comparison: ...
Example of an SLI with reference to the
mz_id filter from the SLO:
spec_version: "1.0" indicators: throughput: "builtin:service.requestCount.total:merge(0):count?scope=tag(keptn_service:$SERVICE),mzId($mz_id)"
By default, Keptn compares the current values with the previous values of the SLIs. To support more advanced comparison strategies, the following properties are available:
compare_with: Defines how many previous results are considered for the comparison: single_result or several_results. (single_result is the default and is used if not specified).
include_result_with_score: Controls which of the previous evaluations are included in the comparison. Therefore, this property accepts the values: pass, pass_or_warn, or all (all is the default and is used if not specified), which target the overall evaluation result and not the single SLI evaluation. In other words, the overall evaluation result decides whether SLI values are considered for the comparison or not.
number_of_comparison_results: Defines the exact number of previous results to consider (1 is the default and is used if not specified).
aggregate_function(optional): Allows overriding the default aggregation function which is
If you configure both of the following:
compare_with: "single_result" number_of_comparison_results
compare_with setting overrides the
Comparison calculations are based on the true
float64 encoded values
but the evaluation results displayed on the Bridge are rounded to two decimal points.
This can lead to surprising results when a value is extremely close to the threshhold.
For example, if the SLO is ≤ 4.00 and the true calculated value is 0.4009601926605503,
the Bridge displays the result as 4.00 but the SLO fails because the true value is slightly more than 4.00.
comparison: compare_with: "single_result" include_result_with_score: "pass" aggregate_function: avg
This comparison configuration means that the current result is only compared to the last result that passed.
comparison: compare_with: "several_results" number_of_comparison_results: 3 include_result_with_score: "pass_or_warn" aggregate_function: "avg"
This comparison configuration means that the current result is compared to the average of the three previous results that had pass or warning as a result.
An objective consists of:
sli: Refers to the name of the SLI.
pass: Represents the upper limit up to which an evaluation is successful.
warning(optional): Describes the border where the result is not pass and not fail, and a manual approval might be needed to decide.
weight(optional): Can be used to emphasize the importance of one SLI over the others. By default,
weightis 1 for all SLIs and can be overwritten. The weight is important for calculating the score later.
key_sli(optional): Can be set to true, meaning that the objective is not met if this SLI fails.
Configuring the criteria:
The pass and warning criteria allow a list of boolean expressions
with a logical operator [<, <=, >, >=] and a absolute or relative value.
While the absolute value is a numerical number,
the relative value requires a (+/-) at the beginning and a % sign at the end, e.g.:
pass: - criteria: - "<1000" - "<=+10%"
pass: - criteria: - "<1000" - criteria: - "<=+10%"
Example of an Objective:
objectives: - sli: response_time_p95 pass: - criteria: - "<=+10%" - "<1000" warning: - criteria: - "<=800" weight: 2 key_sli: true
An evaluation for pass counts for one point, an evaluation for warning half a point, and an evaluation for fail zero points. The default weight of an SLI is 1 and can be overwritten. The maximum score is the sum of the weights of all SLIs.
The actual evaluation result is divided by the maximum score and gives the
total_score in percent.
For example, if the maximum score is 92 and the evaluation result is 85,
total_score is 92.39% (85/92*100).
The pass and warning criteria for the
total_score use the logical operator “>=” by default.
Important: In the following command, the value of the
resourceUri must be set to
To add an SLI configuration to a service, use the keptn add-resource command:
keptn add-resource --project=sockshop --stage=staging --service=carts --resource=slo-quality-gates.yaml --resourceUri=slo.yaml