Module streaming 

Streaming updates for vector indices.

The Problem

Real-world vector databases face continuous updates:

• Documents added as they’re created
• Documents deleted for GDPR/legal compliance
• Embeddings updated when models change

Naive approaches rebuild the entire index, which is expensive. This module provides efficient streaming update primitives.

Architecture

Stream of Updates
    │
    ▼
┌──────────────┐
│ StreamBuffer │ ◄── Batches small updates
└──────┬───────┘
       │
       ▼ (periodic merge)
┌──────────────┐
│  Main Index  │ ◄── HNSW, DiskANN, etc.
└──────────────┘

Update Types

Type	Complexity	Notes
Insert	O(log n) amortized	Buffered, merged periodically
Delete	O(degree)	Mark-and-repair (IP-DiskANN pattern)
Update	Delete + Insert	Atomic replacement

Research Context

• FreshDiskANN (Singh et al. 2021): StreamingMerge for efficient insert/delete

  • Key insight: Maintain a small “fresh” index alongside main index
  • Insertions go to fresh index; periodic merge into main
  • Deletions: tombstone + lazy cleanup during merge

• IP-DiskANN (2025): In-neighbor tracking for O(1) delete preparation

  • Each node tracks its in-neighbors (who points to it)
  • Delete: O(degree) edge repairs instead of O(n) search

• SPFresh (2024): Streaming vector search with LIRE (Lazy In-place REfresh)

  • Lazy updates: defer graph repairs until node is visited
  • Achieves 2-5x throughput over eager updates

• LSM-VEC (2025): LSM-tree approach for streaming vectors

  • Multiple levels of indices, periodic compaction
  • Write-optimized: O(1) amortized insert

Our Approach

We implement a hybrid:

1. Write buffer: Absorbs burst writes (like LSM memtable)
2. Tombstone deletes: Mark-and-sweep during compaction
3. Merged search: Query both buffer and main index
4. Periodic compaction: Merge buffer into main index

Example

use vicinity::streaming::{StreamingIndex, UpdateOp};
use vicinity::hnsw::HNSWIndex;

let mut index = StreamingIndex::new(HNSWIndex::new(128)?);

// Stream of updates
index.apply(UpdateOp::Insert { id: 0, vector: vec![0.1; 128] })?;
index.apply(UpdateOp::Insert { id: 1, vector: vec![0.2; 128] })?;
index.apply(UpdateOp::Delete { id: 0 })?;

// Search works across buffered and merged data
let results = index.search(&query, 10)?;

// Periodic compaction
index.compact()?;

Structs

StreamBuffer

A buffer for streaming vector updates.

StreamBufferConfig

Configuration for the stream buffer.

StreamingConfig

Configuration for streaming coordinator.

StreamingCoordinator

Coordinator that wraps an index with streaming update support.

StreamingStats

Statistics about streaming index state.

UpdateBatch

A batch of update operations.

UpdateStats

Statistics from applying updates.

Enums

UpdateOp

A single update operation.

Traits

IndexOps

Trait for index operations needed by streaming coordinator.

StreamingIndex

Trait for indices that support streaming updates.