Benchmark
Benchmarking real-time graph performance at scale
What you'll learn
Identity resolution at scale exposes the limits of most graph databases fast. Latency spikes under load, infrastructure costs compound unpredictably, and systems that handled millions of edges buckle long before reaching billions. This benchmark documents how Aerospike Graph maintains stable, predictable performance across that entire range.
Using real-world AdTech identity graph workloads, the tests run from 200GB to 20TB of data, measuring query latency, throughput, and cost at every step. The results show what it takes to build graph infrastructure that behaves consistently as scale evolves, not just when conditions are favorable.
Key highlights
Real-world AdTech identity graph workload and data model tested across billions of edges
Query performance stays stable as data scales from 200GB to 20TB with no degradation
600K QPS throughput demonstrated with linear scaling across the full data range
Cost per GB cut in half as data grows, showing predictable infrastructure economics at scale
Zero performance trade-offs as scale evolves, so your deployment plans on solid ground