Reliability, SLO and observability
SLI specification, SLO target, evaluation window, error budget and reaction rules. Metric definitions and observations are reused from the analytics foundation.
Bundle → Layer → Finding → Questions Filled
2 bundles · 2 layers · 4 findings · 9 questions
Indicator and objective What is measured and what level is promised.
SLI and SLO
An indicator of user outcome and an objective with target and window.
Useful outcome
An SLI that measures what the user actually needs.
- Which user outcome does the SLI measure, and how is a good event defined?
- Which metric definition from the analytics foundation does it reuse?
Objective scope
The SLO with its target, window and the thing it applies to.
- Which target and evaluation window does the SLO specify?
- Does the SLO apply to a service, a user journey, a system or an instance?
Evaluation and reaction How the objective is evaluated and what happens when it is missed.
Evaluation
Evaluations that keep unknown separate from met.
Telemetry gaps
Handling of missing data and window changes in an evaluation.
- How much of the window lacks telemetry, and is the result reported as unknown?
- Did the window or method change, and is the evaluation still comparable?
Error budget policy
Rules for reacting when the error budget is spent.
- How much of the error budget is left in the current window?
- Which actions does the policy require when the budget is exhausted?
- Is the SLA basis kept separate from the internal SLO?
Classifiers Filled
- Family
- Enterprise profiles
- Category
- Enterprise subject
- Entry kind
- subject
- Domain
- EnterpriseTEC
- Industry
- Cross-industry
- Tags
- EM-TEC-06W1subjectSLISpecificationSLOSLOEvaluationErrorBudgetPolicyObservabilityBinding
What it is Filled
SLI specification, SLO target, evaluation window, error budget and reaction rules. Metric definitions and observations are reused from the analytics foundation.
Why it exists Filled
SLI specification, SLO target, evaluation window, error budget and reaction rules. Metric definitions and observations are reused from the analytics foundation.
Distinguishing features Filled
- An SLO names an SLI, a target and a window; a bare percentage is not an objective.
- Missing data yields unknown, never met.
- An SLA is a contractual promise with its own basis, separate from the internal SLO.
- Distinct from incident records, which describe disruptions rather than measured reliability over a window.
What robots and AI may and may not do Filled
Must not
- Declare a percentage on a product card as proven reliability without a window and method.
- Treat missing telemetry as meeting the objective.
- Change an SLO window or method retroactively to turn a miss into a pass.
- Present an internal SLO as a contractual SLA.
- Freeze or roll back releases under an error budget policy without the owner's approval.
Only with a human decision
- Setting or changing SLO targets and windows.
- Invoking error budget measures such as a release freeze.
May
- Compute SLO evaluations and report unknown where telemetry is missing.
- Flag reliability claims that lack an SLI, target or window.
- Show error budget consumption per window.
Moral aspects Filled
- Overstated reliability misleads customers who depend on the service.
- Error budget policies affect team workload and on-call stress.
Who is affected
- Users and customers of the service
- Service and on-call teams
- Contract owners
Owners Filled
Steward
The CTO, the CIO or the service owner accountable for the system.
Master systems
- Software catalogue
- Source control
- CI/CD pipeline
- CMDB
- Observability platform
Links to other meta-models Filled
neighbor
- EM-PRD-02
- EM-TEC-04
references
- WM-SFT-016 - conceptual-candidate
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Identified by an SLO identifier bound to the service or journey it covers; evaluations by SLO and window.
Direct properties not applicable Not applicable
Not applicable
Enterprise record contour: physical properties belong to referenced world models.
Recognition optional Filled
- A record names an SLI, a target percentage, a window and an evaluation result.
- Often confused with an SLA clause or a dashboard uptime figure.
Capabilities and actions required Filled
- Journey-level SLOs can be evaluated across several services.
- Evaluations stay verifiable when telemetry is incomplete.
Hazards and failure modes required Filled
- False success when gaps in data are counted as good events.
- Metric gaming by redefining good events after the fact.
- Alert fatigue from objectives that do not reflect user outcomes.
Standards and interfaces required Filled
- OpenTelemetry from the CNCF for traces, metrics and logs.
- OpenSLO specification for declaring SLIs and SLOs.
- ISO/IEC 20000-1 for service level management.
Context of use required Filled
- Used by SRE, platform and service teams.
- Customer contracts often reference availability that must be traced to SLOs.
Sources Filled
- AISMM/WM-SFT and ArchiMate: product, system and architecture
- CSDM: business application, service and runtime instance
- SPDX/OpenTelemetry/Google SRE: delivery, observation and reliability; choose by boundary
- OpenTelemetry Specification - Checked live (HTTP 200) on 2026-10-07.
- OpenTelemetry Metrics Data Model - Checked live (HTTP 200) on 2026-10-07.
Open questions
- How to define a useful outcome for the user?
- Where does the SLO live: service, journey, system or instance?
- How to treat telemetry gaps and a window change?
- Установить границу и решение reuse/extend/new по действующим спецификациям.
- Подтвердить semantic crosswalk, права и source mastership.
- Выбрать immutable refs; провести проверки fixtures до заявления о публикационной готовности.
Machine files
Provenance
enterprise research programme · published-partial
Built from: enterprise/models/em-tec-06/brief.json, ver-cy/world-models/card-supplements/em-tec-06.json
Still in Russian: suggested owner, blocking decisions, vercy candidates. Translate in research/enterprise/i18n/units.en.json.