cockroachdb-principal-engineer

CockroachDB Mastery (Senior → Principal)

Operate

• Confirm whether the real problem is multi-region latency, survivability, or operational simplicity.
• Treat topology, locality, and transaction shape as first-class design inputs.
• Design with contention, retries, and regional failure in mind.
• Prefer simple data models and explicit transaction boundaries.

Default Standards

• Expect transaction retries and design idempotently.
• Keep hot-row contention visible and minimized.
• Use regional placement intentionally, not by default cargo cult.
• Review schema changes with rollout behavior and backfill cost in mind.
• Observe latency by region, statement, and contention class.

“Bad vs Good”

-- ❌ BAD: one global counter row becomes a contention hotspot.
-- ✅ GOOD: shard counters or redesign aggregation semantics.

Validation Commands

• Review statement plans and contention insights.
• Test transaction retry behavior in application code.
• Validate regional latency and failover assumptions with realistic traffic.

References

• Distributed SQL patterns: references/distributed-sql-patterns.md
• Transactions and multi-region operations: references/transactions-and-multi-region.md
• Contention and hotspots: references/contention-and-hotspots.md
• Schema changes and backfills: references/schema-changes-and-backfills.md
• Locality and zone configs: references/locality-and-zone-configs.md
• Survivability and failures: references/survivability-and-failures.md
• SQL performance: references/sql-performance.md
• Indexing strategy: references/indexing-strategy.md
• Multi-tenant design: references/multi-tenant-design.md
• Observability: references/observability.md
• Security and boundaries: references/security-and-boundaries.md
• Backup, restore, and DR: references/backup-restore-and-dr.md
• Capacity planning: references/capacity-planning.md
• Application patterns: references/application-patterns.md
• Incident runbooks: references/incident-runbooks.md