Performance
Ramify DB is designed for speed, operating entirely in-memory. However, as datasets grow, following performance best practices ensures your application remains buttery smooth.
Performance Strategies
1. Strategic Indexing
Indexes are the single most important factor for read performance in where queries.
- Always index fields used in
whereclauses for exact-match filtering (equals,anyOf,allOf). - Note: Indexes do NOT improve sorting. Sorting uses JavaScript's
Array.sort()on the filtered results, regardless of whether the field is indexed. Indexes only optimize the initial filtering step. - Over-indexing Warning: Each index slows down write operations because every add/update/delete must update all indexes. However, memory impact is minimal since indexes store references to documents, not copies.
2. Batch Operations
When inserting or updating multiple records, always use bulk methods (bulkAdd, bulkPut,
bulkDelete).
- Why? Each individual operation triggers index updates and event emissions (for Live Queries). Bulk methods coalesce these into a single event and optimized index build, which is significantly faster.
3. Optimize Live Queries
Live Queries are powerful but come with overhead.
- Granularity: Subscribe to specific datasets rather than whole collections if possible.
- Component Design: Avoid having hundreds of small components each with their own
useLiveQuery. Instead, query data higher up the tree and pass it down via props. - Selective Subscriptions: Subscribing to data that rarely changes still consumes cycles. Only use Live Queries for data that requires reactive updates.
Benchmark
Benchmarked using Benchmark.js for statistically significant results.
Environment
- CPU: Apple M4 (10 cores)
- RAM: 16.00 GB
- OS: darwin arm64
- Runtime: Node.js v22.19.0
- Flags: None
Memory Usage
| Scenario | Heap Used (MB) |
|---|---|
| Empty DB | 12.17 |
| After Load (100k records) | 99.00 |
Results
Dataset: 100k messages
| Operation | ms/op | Ops/sec | ±% | Samples |
|---|---|---|---|---|
| get(id) | 0.000034 | 29,174,761 | 0.65 | 96 |
| bulkGet(50 ids) | 0.004326 | 231,142 | 0.94 | 99 |
| has(id) | 0.000003 | 287,643,315 | 4.99 | 82 |
| count() | 0.000004 | 285,249,233 | 3.95 | 80 |
| where(criteriaObject) | 9.691231 | 103 | 5.22 | 69 |
| where(index).equals() | 9.208464 | 109 | 5.27 | 67 |
| where(index).anyOf() | 29.178278 | 34 | 5.90 | 58 |
| where(multiEntry).allOf() | 11.292184 | 89 | 2.48 | 78 |
| add(doc) | 0.003467 | 288,450 | 12.04 | 70 |
| put(doc) | 0.004037 | 247,684 | 4.45 | 91 |
| update(id, changes) | 0.000031 | 32,547,199 | 0.73 | 98 |
| bulkPut(100 docs) | 0.421761 | 2,371 | 16.81 | 86 |
| bulkUpdate(50 ids) | 0.004415 | 226,477 | 0.25 | 92 |
| where(index).modify() | 28.333660 | 35 | 1.04 | 59 |
| delete(id) | 0.000035 | 28,208,391 | 1.49 | 88 |
| bulkDelete(100 ids) | 0.006415 | 155,881 | 101.79 | 93 |
All benchmarks run until achieving statistical significance.