Is Qdrant Query Speed Inconsistency Due to Caching? #4540
dylanewbie
started this conversation in
General
Replies: 1 comment
-
hey @dylanewbie, if you are using any kind of on-disk option for vectors or indexes, the first request indeed will hit the cold disk cache. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I would like to know if Qdrant has a caching mechanism.
I have a collection with 9 million points and I am using a Rust program with an HTTP client to call the Qdrant API for queries. The program receives a CSV file containing 10,000 points to query.
I noticed that the first query takes 90 seconds (excluding CSV loading time). For the same data, subsequent queries take only 20 seconds. However, when I switch to new data, the first query is slow again, but subsequent queries are fast.
Does Qdrant have a caching mechanism? How can I temporarily disable caching? I need to measure the average query time, but currently, I can only get accurate measurements by restarting the system, as the times after the first query do not seem reliable.
Thank you!
Beta Was this translation helpful? Give feedback.
All reactions