Blog
PostgreSQL, pgvector, and AI infrastructure: guides and deep dives.
Filter by tag
Featured
Crunchy Bridge Alternatives: Managed Postgres Compared
An honest look at Crunchy Bridge: what it does well, what it costs, and when a focused managed-pgvector service is the better fit.
All posts
pgvector vs pgvectorscale: Which You Actually Need
StreamingDiskANN, SBQ, and the RAM line that decides it. When tuned vanilla pgvector wins, and when DiskANN in Postgres earns its keep.
10 min read
pgvector vs Qdrant (2026): Speed, Filtering, Cost, Ops
Head-to-head on query speed, metadata filtering, cost at 1M vectors, and day-2 operations. When Postgres is enough, and when Qdrant earns the extra service.
11 min read
Self-Hosted vs Managed pgvector: The Real Total Cost of Ownership
Self-hosted vs managed pgvector compared on true TCO. Raw compute is cheaper on a VM you run yourself, but pgvector cost is mostly engineer-hours for HNSW tuning, pooling, backups, and HA. An honest breakdown with rough, labeled estimates.
11 min read
Semantic Search, Built Into Your Postgres
Run meaning-based search on your own dedicated PostgreSQL with pgvector and HNSW, no separate vector database. Embeddings generated for you, a one-click sample dataset, and query times measured by Postgres.
4 min read
How to Choose a Managed PostgreSQL Provider Without Getting Burned
A buying guide for managed PostgreSQL from someone who has paid the surprise bills: dedicated vs shared compute, NVMe vs network storage, PITR upsells, benchmark-marketing tricks, EU residency, and a trial-week evaluation checklist.
12 min read
Managed PostgreSQL vs Self-Hosted: What Running Your Own Actually Costs
The real checklist behind self-hosting PostgreSQL (restore-tested backups, failover, patching, disk management): what it costs in hours per month, when self-hosting genuinely wins, and what to demand from any managed provider.
11 min read
HNSW vs IVFFlat in pgvector: How to Actually Choose
A practitioner's decision guide to pgvector's two index types: how HNSW graphs and IVFFlat clusters really behave, measured build times and QPS on real hardware, the ef_search-over-PgBouncer trap, and a decision rule that holds up.
11 min read
PostgreSQL vs Dedicated Vector Databases: Where the Line Actually Is
Pinecone, Qdrant, Weaviate, and Milvus versus Postgres with pgvector: what each side genuinely does better, the real cost of running two systems, and measured numbers showing where a single Postgres node stops.
10 min read