Vector Database Sharding

1. The Pattern

When the similarity index outgrows one machine, split it. But similarity search breaks the best rule from Module 3. A nearest-neighbor query has no key. The closest vectors could live anywhere. So the default is scatter-gather to every shard. The craft is finding the cases where a routable key returns.

First, the arithmetic that forces the split. An embedding is a list of numbers that stands for one piece of content, and each number is one dimension