Week 2 – Storage Architecture & Selection on Google Cloud Platform¶
Focus: Navigating storage trade-offs, vector engines, hybrid databases, and graph RAG on GCP.
Learning Outcomes¶
- Compare Vertex AI Vector Search, AlloyDB / Cloud SQL with pgvector, BigQuery Vector Search, and graph options (Neo4j / Spanner Graph).
- Select storage based on query type (structural/graph vs. semantic policy), latency, cost, and operational complexity.
- Benchmark hybrid (AlloyDB-style) vs pure semantic (Vector Search-style) backends on the Week 1 corpus.
- Build a Graph-Augmented RAG path that detects DFD trust-boundary crossings and attaches governing controls.
Agenda¶
- Theory – Storage options & trade-off matrix
- Lab 2.1 – AlloyDB-style hybrid vs Vector Search-style pure semantic
- Lab 2.2 – Graph-Augmented RAG for DFD connectivity