Persistence and database concepts
These are the topics in this section, each on its own page with a stable path you can bookmark or share.
What topics are covered in this section?
- What is a vector database?
- What is a vector index?
- What is index persistence?
- What is a binary index format?
- What is format versioning?
- What is endianness?
- What is checksum validation?
- What is an atomic snapshot?
- What is a write-ahead log (WAL)?
- What is an incremental checkpoint?
- What is tombstone deletion?
- What is deleted-slot reuse?
- What is a transaction, in a database?
- What is multi-tenancy in a vector database?
- What is resource governance?
- What is query-planner integration?
Persistence and database concepts cover what happens when HNSW lives inside a vector database: durable indexes, WALs, snapshots, tombstones, tenancy, transactions, and query planning.
What does a database add on top of a graph algorithm?
A vector index must persist across restarts via binary formats, versioning, endianness, checksums, atomic snapshots, write-ahead logs, and incremental checkpoints. Tombstone deletion and deleted-slot reuse manage mutability. Transactions, multi-tenancy, resource governance, and query-planner integration situate ANN beside filters, keyword search, and hybrid plans. Without durability and isolation, a fast in-memory graph is only a demo.
These terms explain Weaviate’s job beyond “run HNSW.”
How does Weaviate embody them?
Weaviate pairs object and inverted stores with per-shard vector indexes, commit logs, HNSW snapshots, tombstone cleanup, replication, and RBAC-aware multi-tenancy. Resource limits and async indexing protect clusters under ingest. Understanding WAL and snapshot vocabulary makes restart times and disk growth interpretable. Part IX’s vector-database chapter is the curriculum bridge.
Persistence bugs show up as missing objects, replay failures, or silent corruption – see failure modes.
Where next?
Read vector database, vector index, WAL, tombstone deletion, and multi-tenancy pages, then Part IX and Part IV’s persistence chapter. For cluster healing, see replication and distributed graph repair.
This section places HNSW inside database machinery. Next, open “What is a vector database?” or Part IX’s database chapter, then WAL and snapshot terms when durability is on your checklist.