Langfuse Self Hosting Enterprise vs. FOSS
Hi. i am looking to understand difference between self-hosting (FOSS) and cloud serving (Team). The docs say that SOC2 is not available in self-hosting but why would one need SOC2 if the deployment is private on AWS.
Also is it possible to use custom auth providers for 2FA when deploying via FOSS?
4 Replies
Hi Dex,
Have a look here for an overview of what's included in each version of Langfuse: https://langfuse.com/docs/deployment/feature-overview
re SOC2/ISO reports: you don't need them for running Langfuse but enterprise customers usually want them as part of their due diligence into how we write code, how it is tested etc. Same for DPAs even when self-hosting (e.g. governing data transmitted in the course of support/debugging).
You can use custom Auth providers but you cannot exclusively enforce SSO on a domain.
See more here: https://langfuse.com/docs/deployment/self-host#sso
Self-Hosting Langfuse - Langfuse
Self-host Langfuse in your infrastructure using Docker.
Thanks for the overview. I did look at the docs you shared.
Referring to DPA part, should i say that even if we do self-hosting, langfuse is collecting some of the data?
This is be crucial to understand as we are dealing with sensitive data here
Dear @Clemo. I would appreciate your clarification on this
Hi Dex, please allow us a few hours to respond to questions. We're a small team and working on a bunch of things.
No personal data touches Langfuse if you self-host ([+ opt-out of aggregate telemetry[(https://github.com/langfuse/langfuse/?tab=readme-ov-file#telemetry)).
Some enterprises and companies still want to sign a DPA though. If they end up sending us debug logs or traces or we jump on call to help them, we may be exposed to PII. In those cases, and NDA and a DPA may be required.
In short - you may not need it & no PII will touch us by default if you self-host.
GitHub
GitHub - langfuse/langfuse: 🪢 Open source LLM engineering platform:...
🪢 Open source LLM engineering platform: LLM Observability, metrics, evals, prompt management, playground, datasets. Integrates with LlamaIndex, Langchain, OpenAI SDK, LiteLLM, and more. 🍊YC W23 - ...