WIP Limit
A useful WIP limit makes work and waiting visible. Define the unit and boundary, then examine whether the policy improves flow within available capacity.
-
-
Developer Experience
Developer experience covers feedback loops, cognitive load, and flow state. Combine workflow observations with perceptions to investigate work conditions.
-
Inner Source
Inner source enables contributions across teams. Define guest and host capacity, review, acceptance, support, and the lifecycle of shared software.
-
Developer Advocacy
Developer advocacy helps internal teams discover and use a platform, brings user evidence into decisions, and explains the outcomes and support boundaries.
-
Enabling Constraints
Enabling constraints can reduce repeated decisions and contain failure. Their value depends on workload fit, cost, required controls, and user outcomes.
-
Scaffolding
Generated code, referenced artifacts, and operated services need different update paths. Scaffolding can combine them but does not decide the operating owner.
-
Day 2 Operations
Plan maintenance and retirement alongside creation, with named operators, support commitments and usable paths for changes and recovery.
-
Self-Service Provisioning
Self-service can reduce routine handling. Define what users receive, how partial failures are reconciled and who supports the capability after creation.
-
Service Catalog
A software catalog helps users discover owners and dependencies. Confirm responsibility, track ingestion gaps and connect records to support and workflows.
-
Thinnest Viable Platform
Provide enough capability for users to work effectively. Keep implementation thin where suitable, with explicit operating responsibilities, support, and escalation paths.
-
Kubernetes Distribution
Conformance checks required APIs for a version. Moving a workload also requires verifying its infrastructure references, runtime contract, and permissions.
-
Release Channel
Channel enrollment, installed version, and support coverage are different facts. Plan tests and upgrades using the exact target version and service policy.
-
Version Skew
A kubelet three minors behind can block the next control-plane minor upgrade. That limit does not itself block a patch within the current minor.
-
Network Policy
Ingress-only policy does not change egress isolation. Test allowed and blocked paths, and distinguish Pod deletion from confirmed process termination.
-
Graceful Shutdown
preStop consumes the termination budget before the stop signal. Coordinate withdrawal and draining, and keep retries safe if a response is lost.
-
Readiness Probe
Readiness affects endpoint eligibility without restarting a container. Existing connections, publishing settings and routing delays still matter.
-
Persistent Volume Claim
Claim deletion and volume reclaim are separate decisions. Retain preserves an asset for controlled recovery; it does not replace backups.
-
Label Selector
Selectors identify matching objects; owner references record control relationships. Compare labels, endpoints, and intended membership when diagnosing errors.
-
Control Loop
Continuous reconciliation can coexist with completed rollouts and Jobs. Observe progress, controller health, and service outcomes separately.
-
Desired State
A recorded target needs a working reconciler. Check completion, managed scope, and repeated effects rather than treating acceptance as success.
-
OCI (Open Container Initiative)
Image, runtime, and distribution specifications separate packaging, execution, and registry exchange. Compatibility depends on features and platforms.
-
Container Image
A layer is a changeset, not a snapshot. Removals are recorded as entries, which is why a deleted credential still ships.
-
Container
Isolation is not a property a Linux container has. It is a list of specific things you asked the kernel to hide, one flag at a time.
-
Pipeline Template
Most teams build a template as a convenience and describe it as a control. The difference is whether bypassing it is prevented or merely discouraged.
-
Pipeline as Code
The pipeline becomes reviewable, which is the point. It also becomes branch-specific, which is both the feature and the trap.
-
Build Agent
A build agent supplies pipeline compute and forms part of a trust boundary. Its isolation, credentials, retained state, and lifecycle determine what executed code can reach.
-
Secrets Management
Secrets need a tested lifecycle: creation, rotation, revocation, and expiration, including enforcement at the protected service.
-
Infrastructure State File
Terraform state is an operational record, stored locally or through a backend. It can contain sensitive values, so control access and test recovery of resource bindings.
-
Infrastructure as Code
Use reviewed code changes by default, and reconcile authorized emergency changes with the intended state.
-
Blue-Green Deployment
Switching traffic back is fast. It does not un-write what the new version already wrote, which is the part worth planning for.
-
Expand and Contract
Add a compatible new form, migrate callers, then retire the old form. Set the migration window from consumer needs and the rollback commitment.
-
Feature Flag
Release flags need exit criteria; long-lived controls need owners and review dates. Test the production, fallback, and operational states you support.
-
Reproducible Build
Timestamps, locale, paths, and ordering can change build output. Matching bytes checks reproducibility; it does not authenticate every provenance claim.
-
Environment Promotion
Promotion preserves artifact identity and earlier test records. Data, infrastructure, configuration, and traffic differences still need validation.
-
Build Artifact
Connect the release version to retained content, and record the dependencies and runtime needed to install and verify that artifact.
-
Small Batches
Fixed release costs encourage batching, but delay and recovery risk matter too. Reduce repeated overhead while building decomposition and testing skills.
-
Value Stream Mapping
Bring together the people who perform and receive the work. Use their observations and flow measurements to identify delays and test improvements.
-
DORA Metrics
DORA uses five delivery metrics. Define event boundaries, distinguish failed-deployment recovery from all-cause MTTR, and interpret changes in context.
-
Developer Portal
A developer portal makes discovery and access workable, including automatic low-risk paths and explicit review for sensitive operations.
-
DataOps
Treating analytics as code, versioned and tested. The practices are borrowed from DevOps; the hard part — who owns which number — is not a tooling problem.
-
Semantic Versioning
MAJOR.MINOR.PATCH tells users whether an upgrade can break them — but only if what counts as the public interface was declared first. That applies to data, too.
-
Deprecation
A deprecated interface may keep working without a fixed removal date. Explain the support policy and, when retiring it, provide a migration plan.
-
Platform Engineering
Platform engineering builds an internal product other teams consume, so common capabilities are paved once instead of rebuilt by every team that needs them.
-
Platform as a Product
Understand users, maintain a roadmap, and evaluate task outcomes. Voluntary adoption adds evidence where choice exists; mandated platforms still need product management.
-
Golden Path
A golden path packages supported choices for a common task. State its scope, required controls, and the approval and support agreement for alternatives.
-
Internal Developer Platform
An internal developer platform is what a platform team ships — integrated capabilities offered through self-service interfaces, with defaults, guardrails, and a support commitment.
-
Test Oracle
Define how a test decides correctness, preserve relevant duplicates and tolerances, and avoid deriving expected results from the same flawed logic.
-
Test Fixture
Create controlled input and environment for a test, include boundary cases, and clean up resources without losing the evidence needed to debug.
-
Git Branch and Merge
Understand branches as movable commit references, distinguish fast-forward from a merge commit, and resolve conflicts without assuming correctness.
-
Git Commit and Staging Area
Separate the working tree, staging area, and committed snapshot, and learn which version git diff, git commit, and git restore actually use.