~/rivestack/blog $ _
Blog
// PostgreSQL, pgvector, and AI infrastructure — guides and deep dives.
# filter by tag
# featured
Crunchy Data Alternative: A Managed pgvector Option for AI Teams
An honest look at Crunchy Bridge as a managed Postgres provider, and where a focused, EU-based managed pgvector service like Rivestack fits as an alternative.
# all posts
pgvector vs pgvectorscale: When Vanilla Isn't Enough
pgvector vs pgvectorscale compared honestly — StreamingDiskANN, SBQ, and the RAM line that decides which you need. When tuned vanilla pgvector is enough, and when DiskANN in Postgres earns its keep.
10 min read
pgvector vs Qdrant: Which Should You Use in 2026?
pgvector vs Qdrant compared on performance, filtering, cost, and operations. An honest look at when Postgres is enough and when a dedicated vector engine earns its keep.
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.
11 min read