Limiting RAM in Qdrant #692
Description
I am currently working on optimizing the RAM usage for our Qdrant setup. We have configured our collections to be saved on disk and are using memmap to reduce latency. However, as our data grows, we are experiencing an increase in RAM usage.
Our goal is to limit the RAM usage to some limit lets say 8GB, ensuring that only the most recent collections are kept in RAM, while the least used collections are not. Additionally, we want to avoid any Out-Of-Vocabulary (OOV) errors during this process.
Could you please provide guidance on how to achieve this configuration? Specifically, we are looking for:
- Steps to configure Qdrant to limit RAM usage to 8GB.
- Best practices for implementing a mechanism to keep only the most recent collections in RAM.
- Any additional settings or configurations to prevent OOV errors.
Your assistance in this matter would be greatly appreciated.