What does Temporal cost when self-hosted?

Quick Answer: Self-hosted Temporal is free under the MIT license; the only cost is the infrastructure to run Temporal Server, its persistence layer (Cassandra or PostgreSQL), and optional Elasticsearch for advanced visibility. A small production deployment typically costs $400-$900/month on AWS or GCP as of April 2026.

Self-Hosted Temporal Licensing

Temporal Server and the official SDKs (Go, Java, TypeScript, Python, PHP, .NET) are released under the MIT license. There is no license fee for running Temporal in production, including in commercial SaaS products. This distinguishes Temporal from several other workflow platforms that charge for self-managed usage above a certain scale.

Infrastructure Cost Drivers

The actual cost of running self-hosted Temporal is driven by four components:

Temporal Server Cluster

Temporal Server runs as a stateless service. In production, operators typically run three or more nodes for high availability. Each node runs the four internal services (Frontend, History, Matching, Worker). A production-ready node sized for moderate throughput uses approximately 4 vCPU and 8 GB RAM. On AWS, three m6i.xlarge instances cost roughly $330/month on-demand or $200/month with 1-year reserved pricing as of April 2026.

Persistence Layer

Temporal persists workflow history to either Cassandra or PostgreSQL (MySQL is also supported). For small-to-mid workloads, managed PostgreSQL is often the simplest choice. An AWS RDS PostgreSQL db.m6g.large with 200 GB storage runs approximately $180/month. High-throughput deployments (tens of thousands of workflow starts per minute) typically prefer Cassandra for its write scalability; a three-node Cassandra cluster on i3.2xlarge instances runs roughly $900/month.

Elasticsearch for Visibility

Elasticsearch is optional but recommended for advanced visibility queries in the Temporal Web UI (searching workflows by custom search attributes, for example). A three-node Elasticsearch cluster on AWS OpenSearch t3.medium.search starts around $180/month.

Worker Nodes

Workers run the application code (workflows and activities). They are not part of the Temporal cluster itself; they are regular application servers in the team's existing infrastructure. Worker cost depends entirely on the workload — a typical small deployment might run two worker instances at ~$100/month combined.

Typical Small Deployment Total

Combining the above for a production-ready small deployment (moderate throughput, PostgreSQL persistence, optional Elasticsearch):

Component Monthly Cost (AWS, April 2026)
Temporal Server (3 nodes) $200-$330
PostgreSQL (db.m6g.large, 200 GB) $180
Elasticsearch (3x t3.medium.search) $180
Workers (2 instances) $100
Total $660-$790/month

Without Elasticsearch (standard visibility only), the total falls to approximately $480-$610/month.

Comparison with Temporal Cloud

Temporal Cloud (the managed service from Temporal Technologies) starts at approximately $200/month for a development namespace. Production namespaces are priced by actions executed per month (state transitions) and retained history length. A mid-volume production workload of 1 million actions per month with 7-day retention typically costs $1,500-$3,000/month on Temporal Cloud as of April 2026.

Temporal Cloud includes operational responsibilities that self-hosters must handle themselves: cluster upgrades, persistence scaling, Elasticsearch index lifecycle, SRE on-call, multi-region replication, and SOC 2 compliance posture. For teams without dedicated platform engineers, the managed service often justifies its premium over raw infrastructure costs.

When Self-Hosting Makes Sense

Self-hosting is economical when the team already operates Kubernetes clusters, has in-house SRE capacity, or runs high-volume workloads where Temporal Cloud's per-action pricing exceeds infrastructure cost at steady state. Teams processing hundreds of millions of actions per month frequently find self-hosted Temporal cheaper, even after SRE time is factored in. Smaller teams typically find the managed service worth the premium until workload volume justifies dedicated operational investment.

Related Questions

Last updated: | By Rafal Fila

Related Tools

Related Rankings

Best Open-Source Workflow Engines for Engineers in 2026

A ranked list of the best open-source workflow engines for engineers in 2026. This ranking evaluates code-first workflow orchestration platforms that engineers can self-host, extend, and embed inside existing software stacks. The ranking differs from the broader Best Open-Source Automation 2026 list by focusing specifically on workflow engines intended for developers: platforms that prioritize SDK coverage, durable execution, scalability, and operational controls over visual SaaS-connector automation. It includes durable execution engines (Temporal), data and task orchestrators (Apache Airflow, Prefect), low-code workflow builders with strong self-host stories (n8n, Windmill, Activepieces), and historical agent-based tools (Huginn).

Best Automation Tools for Healthcare in 2026

A ranked list of the best automation tools for healthcare organisations in 2026. This ranking evaluates platforms across HIPAA readiness, audit logging, PHI handling, on-premise or private-cloud deployment options, and integration with clinical and administrative systems. The ranking includes enterprise RPA (UiPath, Automation Anywhere), Microsoft-native automation (Power Automate), general-purpose workflow automation (Zapier on Business tier, Make, n8n self-hosted), and enterprise iPaaS (Boomi). Each entry is evaluated against the specific compliance, data-residency, and clinical-integration requirements that distinguish healthcare from other industries.

Dive Deeper