PostHog Production Readiness Report
All-in-one product analytics, session replay, feature flags, and experiments. Powerful but heavy to self-host.
- Category
- Product Analytics
- Language
- Python / TypeScript
- License
- MIT + paid features
- Deployment
- Kubernetes (Helm)
- Datastore
- ClickHouse + Postgres + Kafka
- Ingest
- JS/Mobile SDKs, API
PostHog bundles analytics, session replay, feature flags, and experiments into one platform. Self-hosting gives you full data ownership, which is the main reason teams take it on.
It is one of the heavier stacks to operate: ClickHouse, Kafka, Postgres, Redis, and object storage all in play. PostHog officially recommends their Cloud for most teams and reserves self-hosting for those with strong infra capability and a clear data-residency requirement.
Production readiness score
Best for
- Teams with strict data-residency requirements
- Organizations with dedicated platform/SRE capacity
- Replacing several tools (analytics + flags + replay) at once
- High event volumes where Cloud pricing is prohibitive
Not ideal for
- Small teams without ops capacity (use PostHog Cloud)
- Quick, low-maintenance analytics needs
- Environments that cannot run Kafka and ClickHouse
This is an expert-level deployment. The ingestion pipeline spans Kafka and ClickHouse, and version upgrades require care across stateful services.
Core analytics store; needs sizing, sharding, and backups.
Event ingestion backbone; a stateful service to operate.
Multi-service migrations; follow Helm upgrade notes closely.
Resource needs grow sharply with event volume.
- Compute
- 8+ vCPU / 16+ GB minimum, scales with volume
- Analytics DB
- ClickHouse (sized to events/day)
- Streaming
- Kafka
- Database
- PostgreSQL
- Cache / Storage
- Redis + S3-compatible object storage
Security considerations
- Keep ClickHouse and Kafka on private networks only.
- Enable SSO/SAML and project-level access controls.
- Configure ingestion to honor data-residency boundaries.
- Apply PII filtering and session-replay masking rules.
Backup & recovery
- Back up ClickHouse with native backups to object storage.
- Back up PostgreSQL for app metadata and configuration.
- Document Kafka topic retention; it is a buffer, not a system of record.
- Rehearse a full restore — multi-store consistency is the hard part.
Monitoring
- Track Kafka consumer lag as the primary ingestion-health signal.
- Monitor ClickHouse merges, disk, and query latency.
- Alert on event-ingestion error rates and dropped events.
- Watch pod resource saturation across the Helm release.
Modest volume, single ClickHouse node
Scaled ClickHouse + Kafka + HA
Sharded ClickHouse, multi-AZ, support
Reference stack
Kubernetes + ClickHouse + Kafka + Postgres + Redis + object storage
- 01IngressLoad balancer + TLS
Event ingestion and app
- 02AppPostHog web + plugins
Dashboards, flags, replay
- 03StreamingKafka
Buffers incoming events
- 04AnalyticsClickHouse
Stores and queries events
- 05DataPostgres + Redis + S3
Metadata, cache, recordings
We deploy PostHog on infrastructure you control.
Skip the trial-and-error. Our team handles architecture, security hardening, backups, monitoring, and upgrades — then hands you a production-ready, documented deployment.
Request PostHog deployment