Why Benchmark
After purchasing a server, running benchmarks helps:
- Verify specs — confirm performance matches advertised
- Detect overselling — identify CPU/disk overselling
- Compare providers — data-driven selection decisions
- Establish baseline — monitor for future degradation
Quick Test (Recommended)
# All-in-one benchmark script
curl -sL yabs.sh | bash
This tests CPU (Geekbench 5/6), disk (fio), and network (speedtest).
CPU Test
# sysbench CPU test (single core)
sysbench cpu run
# Multi-core
sysbench cpu --threads=$(nproc) run
Expected: Modern CPUs score 500-2000 events/sec per core.
Disk Test
# Random 4K read IOPS (database workload)
fio --name=randread --ioengine=libaio --direct=1 --bs=4k --size=2g --numjobs=4 \
--iodepth=64 --rw=randread --group_reporting
Network Test
# iperf3 bandwidth test (server: iperf3 -s)
iperf3 -c <server_ip> -t 30 -P 4
Result Interpretation
| Test | Poor | Fair | Good | Excellent |
|---|---|---|---|---|
| CPU events/sec (1 core) | <300 | 300-800 | 800-1500 | >1500 |
| 4K random read IOPS | <5000 | 5000-20000 | 20000-50000 | >50000 |
| 4K random write IOPS | <3000 | 3000-10000 | 10000-30000 | >30000 |
| Network speed (Mbps) | <100 | 100-500 | 500-1000 | >1000 |
16IDC Tips
- Run yabs.sh immediately after purchase and save results as baseline
- If results are significantly below advertised, request a refund or instance swap
- Run a quick test monthly to monitor for performance degradation
- Disk I/O is the most common VPS bottleneck—4K random read IOPS < 5000 indicates severe overselling