Asopitech VS
Qdrant vs Alopex DB
Qdrant's segments, HNSW and payload indexes, and the scope of shards and Raft. Also explains how it differs from Alopex DB.
Updated: Oct 11, 2026
What Qdrant is
Qdrant is a database specialized for vector search. A point holds a vector and a payload (arbitrary metadata). You operate it through a collection API. Vector indexes and payload indexes are core features.
How it works
- Segments: a collection’s data is divided into segments, each with its own vector and payload storage and indexes. Some segments are appendable, and some allow only reads and deletes.
- Vector storage: vectors sit on disk as memory-mapped files, and you choose how much is held in memory.
- Payload storage: for payload fields you filter on, you create a payload index.
- Change log and versions: a change is first written to a log and numbered, then applied to segments. Each segment and each point records the last version applied, and older changes are ignored. This allows recovery from an abnormal exit using the log.
Main features
Besides vector indexes, there are payload indexes for filtering. The documentation says the former speeds up search and the latter speeds up filtering. Sparse vectors are supported as well as dense ones. Distances include Euclidean, Manhattan, and cosine.
Limits
A collection is divided into one or more shards, and each shard is an independent store of points. Raft covers only cluster topology and collection structure; point operations do not go through consensus. The documentation describes this as a trade of low overhead on point operations for not guaranteeing distributed atomic updates. The number of replicas is set per collection. For high availability, use three or more voting nodes.
How it differs from Alopex DB
Both offer vector search with HNSW.
Qdrant is a server specialized in vector search. Alopex DB brings vector search, KV, and full-text search together with SQL in one engine. You write filters as SQL conditions, so you can combine search with joins and aggregation. It runs as an embedded single-file database.
Alopex DB lets you write vector and attribute conditions in ordinary SQL. Give it a try and check it in the quickstart.