VPS Performance Benchmarks: Real-World CPU, RAM, and Disk I/O Speed Tests

Raw performance numbers separate an adequate VPS from a great one. While providers advertise CPU cores and RAM amounts, real-world benchmarks reveal whether those resources deliver as promised. This guide covers practical benchmarking with industry-standard tools — sysbench, fio, iperf3, and stress-ng — so you can measure your VPS performance with confidence.

Why Benchmark Your VPS

Benchmarking gives you objective data on four critical dimensions: CPU computation throughput, memory bandwidth and latency, disk I/O performance (both sequential and random), and network throughput. Running benchmarks immediately after provisioning establishes a performance baseline. Re-running them quarterly catches resource degradation from provider oversubscription or hardware aging. To compare different providers side by side, view our real-world VPS benchmark comparison tables.

CPU Benchmarks with sysbench

sysbench runs a CPU-intensive prime number calculation and reports the total time and events per second. This isolates raw CPU power without disk or network interference.

Install sysbench:

sudo apt install sysbench
sysbench cpu run

Expected results for common VPS tiers:

VPS Tier	Events/sec (single thread)	Total time (50k primes)
1 vCPU (budget)	1,200 – 1,800	28 – 42s
2 vCPU (mid-range)	2,200 – 3,500	14 – 22s
4 vCPU (high-performance)	4,500 – 7,000	7 – 11s

If your results fall below these ranges, the provider may be overcommitting CPU resources. Check mpstat for steal time to confirm.

Memory Benchmarking

Test memory throughput and latency with sysbench’s memory workload:

# Memory write throughput
sysbench memory --memory-block-size=1M --memory-total-size=10G run

# Memory latency (random access)
sysbench memory --memory-block-size=1K --memory-total-size=1G --memory-access-mode=rnd run

A well-provisioned VPS should achieve 8-12 GB/s sequential throughput and sub-100ns latency for cached operations. Significantly lower numbers suggest memory overcommit or ballooning at the hypervisor level.

Disk I/O Benchmarks with fio

fio is the gold standard for storage benchmarking. Test both sequential and random access patterns:

# Sequential read (bandwidth test)
fio --name=seqread --ioengine=libaio --direct=1 --bs=1M --size=2G --numjobs=1 --iodepth=32 --rw=read

# Random 4K read/write (IOPS test)
fio --name=randrw --ioengine=libaio --direct=1 --bs=4k --size=2G --numjobs=4 --iodepth=64 --rw=randrw --rwmixread=75

Storage Type	Seq Read	4K Random Read IOPS
Local NVMe	1.5 – 3.5 GB/s	50,000 – 150,000
Local SATA SSD	300 – 550 MB/s	15,000 – 40,000
Network Block (Ceph/SAN)	100 – 300 MB/s	3,000 – 10,000

Network Throughput with iperf3

Network performance depends on provider bandwidth allocation and the peering path to your target audience. Test with a public iperf3 server:

# Install iperf3
sudo apt install iperf3

# Run test (10 seconds, 8 parallel streams)
iperf3 -c iperf.he.net -t 10 -P 8

A 1 Gbps port should deliver 700-950 Mbps to a well-connected target. Results under 100 Mbps indicate throttled bandwidth or poor routing. For insights on how network performance varies by provider, compare VPS plans with guaranteed bandwidth on our site.

Creating a Benchmark Report

Save your results to a file for future comparison:

#!/bin/bash
# vps-bench.sh — run all benchmarks and log results
LOGFILE="vps-bench-$(date +%Y%m%d).log"
echo "=== CPU Bench ===" | tee -a $LOGFILE
sysbench cpu run | grep -E "events per second|total time" | tee -a $LOGFILE
echo "=== Disk IOPS (4K random) ===" | tee -a $LOGFILE
fio --name=test --ioengine=libaio --direct=1 --bs=4k --size=1G --numjobs=1 --iodepth=64 --rw=randrw --rwmixread=75 --output-format=json 2>/dev/null | python3 -c "import json,sys; d=json.load(sys.stdin); print(f'Read IOPS: {d["jobs"][0]["read"]["iops"]}, Write IOPS: {d["jobs"][0]["write"]["iops"]}')" | tee -a $LOGFILE
echo "=== Network ===" | tee -a $LOGFILE
iperf3 -c iperf.he.net -t 5 -P 4 | grep "SUM" | tee -a $LOGFILE

Summary

Regular benchmarking ensures your VPS delivers the performance you pay for. Compare your results against provider specifications and re-run benchmarks after any plan upgrade. For objective provider performance comparisons, visit Virtual Servers VPS to see real user benchmarks across different hosting companies.