Databricks provides a managed MCP server for your source. What is the lowest-maintenance path?
Why A is the answer
Configure the managed server
Managed servers reduce maintenance.
In detail
- Managed servers are provided for you.
- Using one means less to run and maintain.
- Build a custom server only when nothing else fits.
Use case
An illustrative scenario, not a real incident.
A team needs MCP access to a Databricks service. A managed server exists, so it configures that rather than building and patching its own.
See it in different ways
flowchart TD
A{"Which MCP server?"} -->|"Databricks provides one"| B["Managed"]
A -->|"Third party with an API key"| C["External, key in Databricks Secrets"]
A -->|"Your own logic"| D["Custom"]sequenceDiagram participant A as Agent participant B as Databricks Secrets participant E as External MCP server A->>B: Read the secret reference B-->>A: Key at run time A->>E: Tool call, authenticated
flowchart TD
Q{"Which server fits?"}
Q -->|"Databricks provides one"| A0["Managed"]
Q -->|"A third party provides one"| A1["External"]
Q -->|"Neither"| A2["Custom"]Key point: Use the option with the least to maintain.
Why the other options are wrong
- B. Build a custom server around it
- A custom wrapper adds code you must maintain.
- C. Cache everything in external tables
- Caching adds a copy that goes stale.
- D. Run your own proxy in front of the source
- A proxy adds another service to run.