hydra janitor command's potential limitation

[surajs1n]

Use-case: Ory Hydra is used in self-hosted mode.

hydra janitor uses the DELETE command internally to clean up the table details. Based on my understanding, the delete command introduces gaps in the DB, which means it doesn't free up space completely, leading to fragmentation: https://stackoverflow.com/questions/1940533/does-sqls-delete-statement-truly-delete-data.

I want to understand, how people ensure this is not happening in the case of self-hosted Ory hydra. Any case study/references.

[vinckr]

Hello @surajs1n

We have created the Ory Enterprise License (OEL) for Ory Hydra to solve this: https://www.ory.sh/docs/self-hosted/oel/quickstart - feel free to reach out to us if you are planning to use Ory Hydra in a business-critical use case.
The OEL is really the only version of Ory that guarantees a clean and scalable solution. For the Ory Open Source version it is provided as is and it is not guaranteed that e.g. hydra janitor works without any issues, especially if you use it in a critical and high load environment.
OpenAI uses it in a high scale environment - see the case study here: https://www.ory.sh/case-studies/openai - technical details are not public though.

[zepatrik]

You're linking to a 15 year old discussion about SQL server, which Hydra does not even support. This kind of cleanup is part of every modern database, there is nothing else Hydra could do than delete the rows. For CockroachDB I know that they have a garbage collection job taking care of this, I'd assume other databases have similar capabilities.