Vector Databases
Store and search embeddings for RAG, semantic search, and similarity applications.
Comparison
| Database |
Best For |
Filtering |
Scale |
Managed Option |
| Chroma |
Local dev, prototyping |
Yes |
< 1M |
No |
| FAISS |
Max speed, GPU, batch |
No |
Billions |
No |
| Qdrant |
Production, hybrid search |
Yes |
Millions |
Yes |
| Pinecone |
Fully managed |
Yes |
Billions |
Yes (only) |
| Weaviate |
Hybrid search, GraphQL |
Yes |
Millions |
Yes |
Chroma
Embedded vector database for prototyping. No server needed.
Strengths: Zero-config, auto-embedding, metadata filtering, persistent storage
Limitations: Not for production scale, single-node only
Key concept: Collections hold documents + embeddings + metadata. Auto-embeds text if no vectors provided.
FAISS (Facebook AI)
Pure vector similarity - no metadata, no filtering, maximum speed.
Index types:
- Flat: Exact search, small datasets (< 10K)
- IVF: Inverted file, medium datasets (10K - 1M)
- HNSW: Graph-based, good recall/speed tradeoff
- PQ: Product quantization, memory efficient for billions
Strengths: Fastest, GPU support, scales to billions
Limitations: No filtering, no metadata, vectors only
Key concept: Choose index based on dataset size. Trade accuracy for speed with approximate search.
Qdrant
Production-ready with rich filtering and hybrid search.
Strengths: Payload filtering, horizontal scaling, cloud option, gRPC API
Limitations: More complex setup than Chroma
Key concept: "Payloads" are metadata attached to vectors. Filter during search, not after.
Index Algorithm Concepts
| Algorithm |
How It Works |
Trade-off |
| Flat |
Compare to every vector |
Perfect recall, slow |
| IVF |
Cluster vectors, search nearby clusters |
Good recall, fast |
| HNSW |
Graph of neighbors |
Best recall/speed ratio |
| PQ |
Compress vectors |
Memory efficient, lower recall |
Decision Guide
| Requirement |
Recommendation |
| Quick prototype |
Chroma |
| Metadata filtering |
Chroma, Qdrant, Pinecone |
| Billions of vectors |
FAISS |
| GPU acceleration |
FAISS |
| Production deployment |
Qdrant or Pinecone |
| Fully managed |
Pinecone |
| On-premise control |
Qdrant, Chroma |
Resources
1---
2name: vector-databases
3description: Use when "vector database", "embedding storage", "similarity search", "semantic search", "Chroma", "ChromaDB", "FAISS", "Qdrant", "RAG retrieval", "k-NN search", "vector index", "HNSW", "IVF"
4version: 1.0.0
5metadata:
6 source: https://skillmd.com/skills/eyadsibai/vector-databases
7---
8
9# Vector Databases
10
11Store and search embeddings for RAG, semantic search, and similarity applications.
12
13## Comparison
14
15| Database | Best For | Filtering | Scale | Managed Option |
16|----------|----------|-----------|-------|----------------|
17| **Chroma** | Local dev, prototyping | Yes | < 1M | No |
18| **FAISS** | Max speed, GPU, batch | No | Billions | No |
19| **Qdrant** | Production, hybrid search | Yes | Millions | Yes |
20| **Pinecone** | Fully managed | Yes | Billions | Yes (only) |
21| **Weaviate** | Hybrid search, GraphQL | Yes | Millions | Yes |
22
23---
24
25## Chroma
26
27Embedded vector database for prototyping. No server needed.
28
29**Strengths**: Zero-config, auto-embedding, metadata filtering, persistent storage
30**Limitations**: Not for production scale, single-node only
31
32**Key concept**: Collections hold documents + embeddings + metadata. Auto-embeds text if no vectors provided.
33
34---
35
36## FAISS (Facebook AI)
37
38Pure vector similarity - no metadata, no filtering, maximum speed.
39
40**Index types:**
41
42- **Flat**: Exact search, small datasets (< 10K)
43- **IVF**: Inverted file, medium datasets (10K - 1M)
44- **HNSW**: Graph-based, good recall/speed tradeoff
45- **PQ**: Product quantization, memory efficient for billions
46
47**Strengths**: Fastest, GPU support, scales to billions
48**Limitations**: No filtering, no metadata, vectors only
49
50**Key concept**: Choose index based on dataset size. Trade accuracy for speed with approximate search.
51
52---
53
54## Qdrant
55
56Production-ready with rich filtering and hybrid search.
57
58**Strengths**: Payload filtering, horizontal scaling, cloud option, gRPC API
59**Limitations**: More complex setup than Chroma
60
61**Key concept**: "Payloads" are metadata attached to vectors. Filter during search, not after.
62
63---
64
65## Index Algorithm Concepts
66
67| Algorithm | How It Works | Trade-off |
68|-----------|--------------|-----------|
69| **Flat** | Compare to every vector | Perfect recall, slow |
70| **IVF** | Cluster vectors, search nearby clusters | Good recall, fast |
71| **HNSW** | Graph of neighbors | Best recall/speed ratio |
72| **PQ** | Compress vectors | Memory efficient, lower recall |
73
74---
75
76## Decision Guide
77
78| Requirement | Recommendation |
79|-------------|----------------|
80| Quick prototype | Chroma |
81| Metadata filtering | Chroma, Qdrant, Pinecone |
82| Billions of vectors | FAISS |
83| GPU acceleration | FAISS |
84| Production deployment | Qdrant or Pinecone |
85| Fully managed | Pinecone |
86| On-premise control | Qdrant, Chroma |
87
88## Resources
89
90- Chroma: <https://docs.trychroma.com>
91- FAISS: <https://github.com/facebookresearch/faiss>
92- Qdrant: <https://qdrant.tech/documentation/>
93- Pinecone: <https://docs.pinecone.io>