Back to Home

DevOps in legacy: monolith and deploys

The article describes the first day of a DevOps engineer in a team with a monolithic core and microservices. Focus on deployment problems, security gates and the need for observability. Key practices: standups, runbook, RBAC automation.

Tame the monolith: DevOps in a real team
Advertisement 728x90

Taming the Monolith: First Days of DevOps in a Legacy System

A new DevOps engineer joins a team where a monolithic app—the heart of the architecture—coexists uneasily with microservices. Shared state between components creates deployment bottlenecks. The pipeline handles builds, tests, staging, and manual approval before production. Issues crop up at access time: permission denied stalls the release ritual.

The team holds daily standups: devs share progress, blockers, and plans. The DBA flags migration risks, the team lead reminds everyone of unwritten rules. Without automation, access approvals turn into a multi-step ordeal: request, training, double approval.

Pipeline Structure and Common Failures

Deployment kicks off with a green pipeline light:

Google AdInline article slot
  • Artifact builds.
  • Unit and integration tests.
  • Staging deployment.
  • Manual approval (the bottleneck).
  • Production deploy.

Security gatekeepers step in at the manual stage: no sign-off means rejection. Rollbacks are routine. The runbook stays in draft, while the legacy wizard (architect) warns against poking the sleeping monolith.

The monolith controls shared state, microservices depend on it, sparking resource conflicts. Observability—metrics, alerts, logs—is set for live deployment during release.

Team Structure and Rituals

The stability squad includes:

Google AdInline article slot
  • Product manager (orchestrates releases).
  • Team lead (manages risks).
  • DBA (database guardian, anti-spontaneous indexes).
  • Developers (fix those FIXMEs in code).
  • Security inquisitor (sign-offs and access).

Standups are brutally honest: yesterday's wins, today's plan, blockers. Agile hides deadlines, but reality (permission denied) wins out. The pager buzzes with timely alerts.

Challenges of Legacy Architecture

The system diagram looks like a dungeon: monolith at the core, microservices on the edges. Shared state leads to:

  • Resource turf wars.
  • Deployment dependencies.
  • Manual interventions.

IaC (infrastructure as code) adoption hits human and process roadblocks. Permissions exist but are chaotic—prime for role and policy refactoring.

Google AdInline article slot

Key Takeaways:

  • Monoliths with microservices need a unified runbook for deploys.
  • Manual approvals are an anti-pattern; automate RBAC.
  • Standups spot blockers early, but security gates stay firm.
  • Shared state breeds incidents; plan the migration.
  • Observability (metrics, alerts, logs) is essential for production releases.

Optimization Roadmap

From day one: standardize the pipeline, cut manual steps, set up observability. Don't touch legacy without a plan—tests first, then release. The team is change-ready if we channel the chaos via Kanban.

— Editorial Team

Advertisement 728x90

Read Next