MVC, MVCC and Causal Trees
summary
Proposes concrete designs from Swarm.js and swarmdb prototypes and experiments showing CRDT metadata often exceeds payload by an order of magnitude and forces different storage tradeoffs. Stores updates as columnar strings into LSMs and relies on LSM merge operators to apply CRDT merges which mostly works but becomes CPU and index bound. Shows MVCC-style versioning and floating branches as practical ways to compress metadata and support snapshots while keeping real time sync.
AI-written summary · about
chapters (9)
- 0:00MVC and MVCC concepts
- 1:47MVC in client-side apps
- 3:53CRDT data storage challenges
- 8:20Adapting databases for CRDTs
- 12:41Indexing strategies
- 15:58Branch databases
- 17:26MVCC limitations
- 21:34Conclusion
- 24:29Future directions
related talks
02h 47mFrom naive to advanced RAG: the complete guide
59mThe Evolution of RAG
01h 03mAgents Everywhere
31mBuild an Agentic RAG system using Langchain, Ollama and Milvus
45mRethinking the Language Runtime for Scale
53mAgents, Tools, and Mcp, Oh My! Next Level AI Concepts for Developers
31mTeaching databases to speak human with LLMs and MCP
29mRAG for Data Engineers: Why Your Metadata Is the Missing Piece