Testing Process

Benchmarks are usually run on a fresh VPS instance with a common Linux distribution such as Debian or Ubuntu. We collect system information, run CPU and disk tests, measure network throughput, record the test date, then summarize the result into structured data for comparison.

Raw benchmark output may still appear in the long-form review article, while normalized values such as price, RAM, storage, Geekbench score, disk throughput, network throughput, and value score are stored in the benchmark collection.

Benchmark Tools

YABS

YABS, or Yet Another Bench Script, is used to collect system details, fio disk results, iperf3 network results, and Geekbench output in a single repeatable report.

Geekbench

Geekbench is used as a CPU-oriented benchmark. We record single-core and multi-core results when available. Single-core scores are useful for latency-sensitive applications, while multi-core scores help compare parallel workloads.

fio

fio is used for disk I/O measurements. We prioritize mixed read/write results and may store 4K random IOPS, larger block read/write throughput, and total throughput when the article includes those values.

Network Speed Tests

Network results may come from speedtest-style tests and iperf3 tests across multiple regions. We separate local or regional results from broader global results when possible, because a VPS can be excellent in one region and weak on distant routes.

How Scores Are Calculated

Scores are normalized summaries, not absolute scientific ratings. In general, CPU score compares Geekbench multi-core performance against a 5,000-point target, disk score compares read throughput against a 2,500 MB/s target, and network score averages download and upload throughput against 5,000 Mbps and 2,500 Mbps targets.

Overall score is a weighted result: 40% CPU, 30% disk, and 30% network, with a small value-for-money adjustment. Value score compares the performance score against normalized monthly USD price. Manual scores in older benchmark data are kept only as fallback when a raw metric is missing.

Data quality badges add context: Verified means the core values are directly available from the review data, Estimated means a value such as currency conversion was normalized manually, and Partial means the benchmark has useful data but one or more measurements need caution.

Pricing Notes

VPS pricing changes frequently. Providers may change plan specs, discounts, renewal prices, stock availability, or currency conversion rates after a benchmark is published.

When a provider advertises a promo price, we try to distinguish it from the recurring price. A one-time discount can make a VPS look cheaper than it really is over time, so comparison data should prefer the recurring monthly equivalent when available.

Transparency

Benchmark entries may include transparency fields for whether the VPS was personally purchased, sponsored, or connected to an affiliate link. Most tests are intended to reflect real customer conditions.

Some WakServer links are affiliate links. If you buy through those links, WakServer may earn a commission at no extra cost to you. Affiliate links do not change the benchmark numbers.

If a server is sponsored or temporarily provided by a host, that should be disclosed in the benchmark data or review notes.

Benchmark Limitations

VPS performance can vary because most VPS plans run on shared infrastructure. Shared CPU allocation, noisy neighbors, host node load, storage contention, provider routing, and maintenance windows can all change the result.

Network tests are especially sensitive to route quality, test server location, congestion, peering, and time of day. A result from Jakarta, Singapore, Frankfurt, or Seattle may not match what another user sees from a different region.

Each benchmark should be treated as a snapshot from a specific test date, location, plan, and provider configuration.