Vector DB Categories: Cloud vs Self-Hosted
What you'll learn: How managed cloud vector databases differ from self-hosted open-source options and when to choose each approach.
The Two Deployment Paths
When you've decided you need a vector database (recall lesson 250), your next choice is where and how to run it. Vector databases come in two main categories:
- Cloud-managed services (e.g., Pinecone, Weaviate Cloud, Qdrant Cloud)
- Self-hosted open-source (e.g., Weaviate, Qdrant, Milvus running on your infrastructure)
Think of this like choosing between a fully-furnished apartment (cloud) versus buying and maintaining your own house (self-hosted).
Cloud-Managed Vector Databases
Advantages:
- Zero infrastructure management—just connect and use
- Automatic scaling and high availability
- Built-in monitoring and backups
- Fast time-to-production (minutes, not weeks)
Trade-offs:
- Recurring subscription costs (can grow with usage)
- Less control over configuration and tuning
- Data residency constraints (your vectors live on their servers)
- Potential vendor lock-in
Self-Hosted Open-Source
Advantages:
- Full control over performance tuning (indexing strategies from lessons 253-264)
- Data stays in your infrastructure
- Lower long-term costs at scale
- Customization freedom
Trade-offs:
- You manage servers, updates, scaling, backups
- Requires DevOps expertise
- Longer setup time
- You're responsible for uptime and disaster recovery
Making the Choice
Start with cloud-managed if you're prototyping, have small-to-medium scale needs, or lack infrastructure resources. Choose self-hosted when you have strict data governance requirements, massive scale, existing infrastructure, or need custom optimizations (like specific index configurations from lesson 264).
Key Takeaway: Cloud vector databases offer convenience and speed at recurring cost, while self-hosted options provide control and long-term cost savings but require infrastructure expertise—choose based on your team's capabilities and constraints.