Yes, an external database cluster can back multiple LangSmith instances, but each instance must use its own isolated database. See below for details and related services (PostgreSQL, ClickHouse, Redis, and S3).

Context

Teams deploying LangSmith on Kubernetes or other environments often want to centralize infrastructure by pointing multiple LangSmith clusters at the same external database engine (for example, a managed PostgreSQL or ClickHouse cluster), and to understand who creates databases and how to isolate data. They also want guidance on Redis and S3 reuse across instances.

Answer

Key principles

• Each self-hosted LangSmith instance requires its own dedicated database for proper schema and data isolation.

• LangSmith does not create the database itself. You must provide a pre-provisioned, empty database; LangSmith manages schemas (tables, indexes) inside that database.

• You may share a single database engine/cluster (e.g., one PostgreSQL or ClickHouse cluster) across many LangSmith instances, but not the same database within that engine.

Service-by-service guidance

• PostgreSQL: Use one engine/cluster for many instances, but create a separate database per instance.

• ClickHouse: Same approach as PostgreSQL—separate database per LangSmith instance.

• Redis: Use a separate Redis installation per instance (Redis databases/DB indexes are weakly isolated and not sufficient).

• S3 (blob storage): Use separate buckets per instance to isolate data.

High-level setup steps

1. Provision an external PostgreSQL cluster (or use a managed service). For each LangSmith instance:

   • Create an empty database with its own credentials.

   • Provide the connection string to the instance (via your deployment values/environment variables per the docs).

2. Provision an external ClickHouse cluster if used. For each instance:

   • Create an empty ClickHouse database.

   • Configure the instance with its ClickHouse DSN.

3. Deploy a separate Redis installation per instance. Configure each instance with its own Redis URL.

4. Create a separate S3 bucket per instance and configure access keys and bucket name for each.

Sources

• External PostgreSQL: https://docs.langchain.com/langsmith/self-host-external-postgres

• Kubernetes databases: https://docs.langchain.com/langsmith/kubernetes#databases

• External Redis: https://docs.langchain.com/langsmith/self-host-external-redis

• External ClickHouse: https://docs.langchain.com/langsmith/self-host-external-clickhouse

• Blob storage (S3): https://docs.langchain.com/langsmith/self-host-blob-storage