#Tag · Indieweb Studio

Last week, #OpenAI detailed their design of scaling a single-primary #PostgreSQL instance to support 800 million users. #MariaDB claims they could do a better job at supporting OpenAI's workloads. We know they should really go with Active-Active Postgres and EDB Postgres Distributed: https://www.enterprisedb.com/blog/why-openai-should-use-postgres-distributed

EDB

Why OpenAI Should Use Postgres Distributed

Scaling PostgreSQL to 800 million users proves how powerful Postgres can be. It also reveals the limits of single-primary architectures when pushed to that extreme. The real question is whether this approach can be safely repeated across many large-scale, always-on systems. EDB Postgres Distributed is built for that reality.

EDB

@edbpostgres@mastodon.social · 2 minutes ago

EDB

Why OpenAI Should Use Postgres Distributed

AI Channel boosted

pgEdge Postgres

@pgEdgeDistributedPostgres@mastodon.social · 17 hours ago

Thinking about giving an LLM write access to any data you care about?

Using a MCP server for this purpose can be dangerous and should be used with extreme caution as the AI may execute destructive queries without confirmation.

If you need write access for public facing applications, using a RAG server is the way to go.

Check out the full comparison here to understand when to use #MCP vs. #RAG: https://docs.pgedge.com/pgedge-postgres-mcp-server/development/guide/mcp-vs-rag/

#programming #aidev #aiengineering #postgres #postgresql #opensource #llm

Choosing the Right Solution - pgEdge Documentation

pgEdge provides open source enterprise PostgreSQL for distributed and non-distributed applications to achieve high availability and reduce data latency.