Metadata index size

@Cory_Pinecone Is there any way by I can find the index size created with the metadata unique keys?

Hi @abhishekjain. Not directly but you can infer it from describe_index_stats(). That’ll show you the pod fullness percentage as well as the number of vectors currently present. If you use our rule of thumb of a single p1 pod holding 500k vectors with 768 dimensions and that a free index can hold 100k vectors with 1536 dimensions, you can get a rough idea of how much your metadata is using up.

For instance, if you’re using a free index and using 1536 dimensions with 50k vectors, if those were raw without indexed metadata, you would expect your pod fullness percentage to be about 50%. If it’s 75%, you can infer that your metadata is taking up half as much space as the vectors themselves.

This isn’t perfect, but it can give you a rough idea of where you’re at with your metadata. We’re making some adjustments to how storage is managed, so this is a temporary workaround until those new features are available sometime in the future.

1 Like

@Cory_Pinecone do vector dimensions effect the storage capacity? Suppose if we’re storing 50 k vectors of dimension 768 and 50k vectors of 1536 in too different indexes. do they acquire same space or it will be different?

@Cory_Pinecone Can you please tell me how do we know which dimension vector covers how much space? Suppose if s1.x1 pod has the capacity to store 5M vectors of 768 dimensions mentioned in the standard plan then if I want to store the 1536 dimension vectors then how much vectors I can store in the same s1.x1 pod?