I am creating a vector DB with 10000 vectors, with each vector of size 300. Assuming an fp16 precision, I should have a db file of size somewhere around ~6MB, assuming overheads for inverted index, log files etc, the total size on disk should be 7 to 8 MB, but I am surprisingly having a size of 750 MB, am I missing something?
hi @shiva_rama !! Welcome to our community
There are a lot of variables that can influentiate on the disk size of a collection. Like efconstruction
, compaction, how your collection is modeled, etc.
Let me know if this helps.
I don’t believe we cover disk in the resource planning:
THanks!