Table 3 Average latency and relative overhead for each platform on a single node
From: Object-NoSQL Database Mappers: a benchmark study on the performance overhead
| Â | Write | Read in-memory | Read on-disk | Update in-memory | Update on-disk | |||||
|---|---|---|---|---|---|---|---|---|---|---|
Samples | n=20.000.000 | n=45.000 | n=750.000 | n=39.000 | n=750.000 | |||||
Platform | Latency (μ s) | Latency (μ s) | Latency (ms) | Latency (μ s) | Latency (ms) | |||||
MongoDB | 403±110 | - | 217±34 | - | 15.9±5.2 | - | 298±106 | - | 19.3±9.1 | - |
EclipseLink | 446±105 | 10.8% | 232±41 | 6.7% | 14.2±5.0 | −10.45% | 579±91 | 93.9% | 16.9±8.0 | −12.0% |
Kundera | 442±96 | 9.9% | 256±57 | 17.7% | 17.1±5.6 | +8.0% | 338±56 | 13.3% | 20.7±9.8 | +7.6% |
Hibernate OGM | 452±72 | 12.3% | 289±42 | 32.8% | 15.1±6.5 | −4.7% | 620±53 | 107.6% | 16.8±8.0 | −12.8% |
Apache Gora | 495±92 | 22.9% | 282±65 | 29.8% | 14.5±5.0 | −8.5% | 570±108 | 91.0% | 17.4±8.2 | −9.5% |
DataNucleus | 550±76 | 36.5% | 309±64 | 42.2% | 14.3±5.0 | −9.8% | 882±49 | 194.8% | 17.7±8.3 | −8.0% |