Triple Store Sizing¤
The following table provides a generic sizing orientation when planning the triple store hardware.
Sizing Factors¤
The actual hardware requirements depend on several factors, including:
- Data Model Complexity: The number of statements and the complexity of the ontology.
- Workload: The types of queries (e.g., analytical vs. transactional) and update frequency.
- Users: The number of concurrent users and their usage patterns.
- Hardware: The specific performance characteristics of the server hardware (CPU, Disk, RAM).
Sizing Table¤
| Statements | RAM | Disk Usage |
|---|---|---|
| ~130 M | 8 GB | ~15 GB |
| ~280 M | 16 GB | ~32 GB |
| ~1.100 M | 32 GB | ~110 GB |
| ~2.500 M | 64 GB | ~290 GB |
| ~20.000 M | 128 GB | ~2.000 GB |
Performance Considerations¤
When planning your infrastructure, consider the following performance priorities:
- CPU: Single-core performance is critical for query execution speed.
- Disk I/O: High IOPS and throughput are essential for load and indexing performance. SSDs are strongly recommended.
- RAM: While CPU and Disk influence performance, RAM is the mostly limiting factor in terms of the maximum amount of triples that can be loaded and queried efficiently.