Description
- Design, deploy, and operate zone-redundant high availability on Azure Database for PostgreSQL flexible server: synchronous standby in a second availability zone, automatic failover in 60120 seconds with zero data loss, and the 99.99% uptime SLA that configuration carries.
- Decide between zone-redundant and same-zone (zonal) HA per workload based on latency tolerance and zone-failure risk, and document the trade-off.
- Run and document planned, unplanned, and forced failover tests at least quarterly; measure downtime from the application side, not from the service side.
- Size max_replication_slots and max_wal_senders correctly (four reserved for HA plus one per read replica and per logical slot) so failovers never fail for lack of slots.
- Build equivalent active-passive clusters on self-managed Azure VMs using streaming replication with Patroni and etcd (Linux) or a supported cluster manager on Windows, fronted by HAProxy, PgBouncer, or an Azure internal load balancer.
- Design cross-region disaster recovery: geo-redundant backups, cross-region read replicas, and promotion runbooks with tested RPO/RTO targets.
- Preserve logical replication across failover: pg_failover_slots on PostgreSQL 16 and earlier, native sync_replication_slots = on with hot_standby_feedback = on on PostgreSQL 17 and later.
Active-active (bi-directional / multi-master):
- Design and operate active-active topologies across Azure regions using the pglogical extension or native logical replication, following Microsoft's published bi-directional replication guidance for Azure Database for PostgreSQL.
- Define and enforce conflict-avoidance and conflict-resolution strategy: partitioned write ownership, non-overlapping sequence ranges or UUIDv7 keys, last-writer-wins rules where appropriate, and monitoring for conflict counts.
- Manage DDL propagation across nodes (pglogical.replicate_ddl_command() or controlled DDL rollout for native replication) and sequence synchronization (PostgreSQL 19 adds native sequence replication via ALL SEQUENCES and REFRESH SEQUENCES).
- Monitor replication lag, slot retention, and retain_dead_tuples settings on every node; alert before WAL retention threatens primary disk.
- Own the decision of when active-active is and is not appropriate, and push back on it where a well-run active-passive design meets the requirement at lower risk.
2. Azure platform administration (critical)
- Provision and manage Azure Database for PostgreSQL flexible server via the portal, Azure CLI, Az.PostgreSQLFlexibleServer PowerShell module, ARM/Bicep, Terraform, and REST API.
- Configure networking: VNet integration, private endpoints, private DNS zones, firewall rules, and migration from VNet-injected to private-endpoint servers (now supported for HA-enabled servers).
- Implement Microsoft Entra ID authentication, Entra group login, managed identities, customer-managed keys (including cross-tenant CMK), and SCRAM-SHA-256 as the password default.
- Manage compute and storage tiers, including Premium SSD v2 (GA with HA, read replicas, geo-DR, and major version upgrade support) and online storage scaling.
- Operate built-in PgBouncer (currently 1.25.x) for connection pooling and tune pool modes per application.
- Run in-place major version upgrades on flexible server (PostgreSQL 18 upgrade GA June 2026), including pre-upgrade extension and large-object checks.
- Manage read replicas (in-region, cross-region, cascading), promote them when required, and keep replica lag within SLA.
- Configure backup retention, long-term retention via Azure Backup, point-in-time restore, and recovery of deleted servers; test restores on a schedule.
- Evaluate and deploy elastic clusters (sharded flexible server) for horizontally scaled workloads where appropriate.
- Set custom maintenance windows and coordinate Azure-initiated minor version patching with application owners.
3. Self-managed PostgreSQL on Linux and Windows (critical)
- Install, patch, and tune PostgreSQL 1519 on RHEL 8/9, CentOS Stream, and Windows Server 2019/2022 running as Azure VMs.
- On Linux: manage systemd units, SELinux contexts, firewalld, LVM/managed-disk layout, kernel and huge-page settings, and PGDG repository packaging.
- On Windows: manage the PostgreSQL service, NTFS permissions, Windows Firewall, scheduled tasks for maintenance, and Windows-specific installer and upgrade behavior.
- Operate backup tooling: pgBackRest or Barman on Linux, pg_basebackup and WAL archiving to Azure Blob Storage on both platforms, with incremental backups where supported.
4. Performance engineering
- Diagnose and tune queries using EXPLAIN (ANALYZE, BUFFERS, WAL, IO), pg_stat_statements, auto_explain, and Azure Query Performance Insight.
- Tune autovacuum per table, including PostgreSQL 19 parallel autovacuum (autovacuum_max_parallel_workers) and the new prioritization scoring, and prevent transaction-ID wraparound.
- Reclaim bloat without downtime using REPACK (CONCURRENTLY) on PostgreSQL 19, pg_repack on earlier versions, and VACUUM FULL only as a last resort.
- Tune asynchronous I/O (io_method, worker counts) introduced in PostgreSQL 18, memory parameters, checkpointing, and WAL compression.
- Design partitioning, indexing (B-tree, BRIN, GIN, GiST, skip-scan on PostgreSQL 18), and extended statistics strategies.
- Use pg_plan_advice and pg_stash_advice on PostgreSQL 19 to stabilize problem plans without permanent hints.
5. Security and compliance
- Enforce least-privilege roles, row-level security, and predefined roles (pg_read_all_data, pg_write_all_data); retire MD5 passwords before PostgreSQL 19 begins warning on them.
- Manage TLS, SNI (new in PostgreSQL 19), certificate rotation, and encryption at rest with customer-managed keys.
- Produce audit evidence with pgaudit and Azure Monitor diagnostic logs for SOC 2, HIPAA, or PCI as required.
- Respond to CVEs within the patching SLA and track Azure-applied minor version patches.
6. Monitoring and incident response
- Build dashboards and alerts in Azure Monitor, Log Analytics, and the Azure HA health status monitor; supplement with Prometheus/Grafana or Datadog for VM-hosted clusters.
- Instrument the new PostgreSQL 19 views (pg_stat_recovery, pg_stat_lock, pg_stat_autovacuum_scores) and the per-process-type log_min_messages.
- Lead Sev-1 database incidents, write blameless post-mortems, and drive the resulting fixes to closure.
7. Migrations and upgrades
- Plan and execute migrations into Azure from on-premises or other clouds using Azure Database Migration Service, pg_dump/pg_restore, logical replication, or pg_createsubscriber.
- Lead major version upgrades on both flexible server and VMs, with rollback plans and application regression testing.
- Replace deprecated features ahead of removal: MD5 auth, RADIUS (removed in 19), standard_conforming_strings = off.
8. Automation and engineering practice
- Keep all database infrastructure in code (Terraform, Bicep, Ansible, PowerShell) with CI-tested changes.
- Write runbooks for every failover, restore, and upgrade procedure and keep them current.
- Mentor junior DBAs and developers on PostgreSQL best practices and review schema changes in pull requests.
Required qualifications
All items in this section are must-haves. Azure hosting experience is non-negotiable: candidates whose PostgreSQL experience is only on-premises or on another cloud will not be considered.
Experience
- 7+ years administering PostgreSQL in production, with at least 3 years on PostgreSQL 14 or later.
- 3+ years hosting and operating PostgreSQL in Microsoft Azure, including at least 2 years on Azure Database for PostgreSQL flexible server (not Single Server, which was retired in March 2025).
- Demonstrated ownership of a production active-passive HA cluster, including at least one real (not test) failover.
- Demonstrated design or operation of a production active-active / bi-directional replication topology, with a clear explanation of how conflicts were avoided or resolved.
- Administered PostgreSQL on both Linux (RHEL or CentOS) and Windows Server in production.
Azure
- Azure Database for PostgreSQL flexible server: HA modes, read replicas, PITR and geo-restore, major version upgrade, server parameters, extensions allow-list, PgBouncer, maintenance windows.
- Azure networking for databases: VNet integration, private endpoints, private DNS, NSGs.
- Microsoft Entra ID authentication for PostgreSQL and managed identities.
- Azure Monitor, Log Analytics (KQL), and Azure Advisor for PostgreSQL.
- Infrastructure as code for Azure: Terraform or Bicep, plus Azure CLI and PowerShell.
PostgreSQL
- Streaming replication, replication slots, logical replication (publications and subscriptions), and pglogical.
- Patroni or an equivalent cluster manager; etcd or Consul; HAProxy or PgBouncer.
- Backup and recovery with pgBackRest or Barman, WAL archiving to object storage, and incremental backups.
- Deep query tuning: reading execution plans, index design, statistics, partitioning, autovacuum tuning, and bloat management.
- Major version upgrades via pg_upgrade, logical replication, and pg_createsubscriber.
- Working knowledge of PostgreSQL 18 features (asynchronous I/O, UUIDv7, virtual generated columns, temporal constraints, OAuth, data checksums on by default) and awareness of what changes in PostgreSQL 19.
Operating systems and tooling
- RHEL/CentOS: systemd, SELinux, LVM, package management from the PGDG repository, shell scripting (bash).
- Windows Server: service management, NTFS permissions, PowerShell scripting, Windows-specific PostgreSQL installation and upgrade behavior.
- Git-based workflows, CI/CD pipelines, and configuration management (Ansible preferred).
Professional
- Experience leading Sev-1 incident response and writing post-mortems.
- Clear written communication: runbooks, architecture decision records, and change requests that non-DBAs can follow.
- Ability to participate in a 24x7 on-call rotation.
Preferred qualifications
None of these are required, but each one strengthens a candidate's case.
- Microsoft Certified: Azure Database Administrator Associate (DP-300) or Azure Solutions Architect Expert (AZ-305).
- Red Hat Certified System Administrator (RHCSA) or equivalent Linux certification.
- Hands-on testing of PostgreSQL 19 betas or release candidates, especially REPACK CONCURRENTLY, parallel autovacuum, WAIT FOR LSN, and sequence replication.
- Experience with Azure Database for PostgreSQL elastic clusters (Citus-based sharding) or with Citus on self-managed clusters.
- Experience with EDB Postgres Distributed, pgEdge, or another commercial multi-master product, with a clear view of what native tooling can and cannot replace.
- Experience with Azure Database Migration Service for online migrations from Oracle, SQL Server, or other PostgreSQL sources.
- Experience operating pgvector, pg_diskann, or Apache AGE on flexible server for AI or graph workloads.
- Kubernetes-hosted PostgreSQL on AKS using CloudNativePG, Crunchy Postgres Operator, or Zalando Postgres Operator.
- Contributions to the PostgreSQL community: patches, extensions, conference talks, or documentation.
- Familiarity with Oracle, SQL Server, or MySQL for cross-platform migration work.
- Security and compliance background: SOC 2, HIPAA, PCI DSS, or FedRAMP audits involving database controls