Posts

Limit Work in Progress Before Adding Delivery Capacity

-
When delivery slows, the default leadership response is often to add people, escalate priorities, or demand tighter estimates. These actions may increase activity, but they rarely improve flow. The more common problem is excessive work in progress. Teams are starting more than they can finish. Engineers split attention across features, incidents, migrations, security findings, and platform upgrades. Each item moves slowly because everything is competing for the same constrained capacity. Busy Teams Can Still Have Poor Flow High utilization looks efficient on a planning spreadsheet. In practice, it creates queues. Code waits for review. Releases wait for validation. Infrastructure changes wait for specialists. Decisions wait for overloaded leaders. These delays are easy to overlook because no single task appears blocked for long. The accumulated waiting time, however, can dominate the delivery cycle. Adding another initiative makes the problem worse. It introduces more coordination...
Post a Comment
Read more

Automate Rollbacks Around User Impact

-
Deployment automation is incomplete if it can move software into production but cannot respond when that software causes harm. Many delivery pipelines treat rollback as a manual judgment call. The deployment is automated, monitoring detects a problem, and then people gather to decide what to do. That delay turns a routine release failure into a customer-facing incident. Define the decision before the release Teams should agree on rollback conditions while preparing the change, not while diagnosing its consequences. The right signals depend on the service, but they should represent user impact rather than internal activity alone. Useful conditions may include: A sustained increase in failed user requests Latency exceeding a service objective A critical business transaction failing Resource saturation that threatens service stability A single noisy metric should not trigger constant reversals. Use a defined evaluation window, minimum traffic volume, and multiple signals ...
Post a Comment
Read more

Put an Interruption Budget on Engineering Teams

-
Engineering leaders often plan capacity as if teams spend every week executing roadmap work. The plan accounts for features, migrations, and technical debt, but treats interruptions as random noise. They are not random. Production support, security requests, dependency upgrades, access issues, and urgent questions create a recurring demand on engineering capacity. When that demand remains invisible, teams appear slow and commitments become unreliable. Measure the demand before fixing it An interruption budget makes this work explicit. It is not a target for how many disruptions a team should tolerate. It is a planning mechanism that shows how much capacity is being consumed outside the intended backlog. Start with lightweight categories rather than detailed time tracking: Production incidents and support Security and compliance requests Developer environment or platform issues Unplanned dependency and infrastructure work Cross-team assistance Review the pattern every few we...
Post a Comment
Read more

Operational Readiness Is a Release Requirement

-
A feature is not ready for production simply because the code works. It is ready when the team can detect failure, understand impact, and recover safely. Too many release processes treat operational readiness as follow-up work. Dashboards, alerts, runbooks, capacity limits, and rollback procedures are deferred until after launch. This creates an avoidable gap: the service is delivering value, but the organization is not prepared to operate it. Define readiness in observable terms “Production ready” should not depend on confidence or seniority. Leaders should establish a small set of verifiable conditions that apply to every meaningful service change. Key user outcomes are measurable. Failure conditions generate actionable alerts. Ownership and escalation paths are clear. Rollback or mitigation steps have been tested. Capacity assumptions and dependencies are documented. This is not a demand for exhaustive documentation. The goal is enough operational context for someone...
Post a Comment
Read more

Make Service Ownership Machine-Readable

-
Most organizations document service ownership somewhere. The problem is that “somewhere” often means a wiki page, spreadsheet, or repository file that becomes stale as teams and systems change. Ownership data should not depend on someone remembering to update a document. It should be structured, versioned, and connected to the systems engineers use every day. Ownership Is Operational Data A service catalog entry should identify more than a team name. At minimum, it should define the responsible group, source repository, deployment pipeline, production environment, support channel, escalation path, and links to operational guidance. This information must be machine-readable. A consistent metadata file or catalog API allows delivery pipelines, observability tools, incident systems, and security scanners to use the same ownership record. When a production alert fires, responders should not search for the team that owns the service. When a vulnerability is detected, security shou...
Post a Comment
Read more

Design Degraded Modes Before the Next Incident

-
Most reliability plans focus on keeping every capability available. That is an admirable goal, but it is not a complete operating strategy. Complex systems will eventually lose a dependency, exhaust a resource, or encounter demand they cannot serve normally. When that happens, teams need more than alerts and runbooks. They need an agreed way for the service to become less capable without becoming completely unavailable. Degradation Is a Product Decision A degraded mode defines which capabilities can be limited, delayed, simplified, or disabled to protect the service’s most important outcomes. An online service might pause recommendations while preserving core transactions. An internal developer platform might delay nonessential analytics while continuing deployments. A data pipeline might prioritize current operational events over historical backfills. Engineering cannot make these tradeoffs alone during an incident. Product, operations, security, and business leaders need to deci...
Post a Comment
Read more

Put Expiration Dates on Your Golden Paths

-
A paved road can become a legacy road Golden paths help engineering teams start quickly and operate consistently. They package sensible defaults for repositories, pipelines, observability, security controls, and runtime configuration. But a golden path is not automatically a current path. Without deliberate maintenance, the recommended approach slowly accumulates outdated dependencies, unnecessary controls, and assumptions that no longer match how teams build software. Because the path carries an official label, developers may continue using it long after better options exist. Platform leaders should treat every golden path as a versioned product with an expiration date. Expiration does not mean automatic deletion. It creates a decision point where the organization must renew, replace, or retire the path. Define renewal criteria before launch A golden path should not remain recommended simply because it still works. Its owners should periodically verify that it remains supportabl...
Post a Comment
Read more