valkey-bundle: One stop shop for real-time applications
summary
Valkey-bundle combines an in‑memory core with JSON path queries, Bloom filters, vector similarity search, and LDAP modules. Demo is a Flask app using hashes for product records and JSON for users, storing 384‑dim embeddings and an HNSW index via the FT command to return 25 nearest neighbors in sub‑10ms. Bloom filters track ~10k seen products per user and claim ~99% memory reduction; LLM descriptions (TinyLlama) cache cuts generation latency from ~2s to <1ms and avoids repeated token costs.
AI-written summary · about
chapters (15)
- 0:00Introduction to Valkey-bundle
- 1:51Demo of real-time personalization
- 5:10Architecture overview of the app
- 6:37Using Valkey hashes for products
- 9:45Valkey JSON for nested data
- 15:18Generating user embeddings
- 16:01Bloom filters explained
- 21:23Vector similarity search
- 27:59Code walkthrough of search logic
- 33:00Personalization pipeline details
- 35:53Session storage and caching
- 40:07Benefits of Valkey-bundle
- 43:59Local development setup
- 44:22Use cases and limitations
- 45:04Q&A session
related talks
36mWhat’s New in Valkey
06mKeynote: Cut Database Costs, Fund Innovation with Valkey
27mAn Introduction to Messaging in Valkey
10mThe unreasonable effectiveness of key value databases
33mDiscover Valkey: The Path to Now
32mKeyvi the Key Value Index Optimized for Size and Speed #bbuzz
48mValkey clusters unboxed
01h 00mHyper Speed: When Big Data Blooms