Who runs HostingSelector

The person

I’m Arnas Kazlaus — Software engineer and founder with 15 years of experience shipping production web apps. I run the tests, write the editorials, and pay the hosting bills. You can reach me at iam@arn.as or at arn.as.

Why this site exists

Every “best VPS for X” article on the internet is written by someone who has never rented that VPS. They regurgitate spec sheets, they copy marketing copy, they rank providers by who pays the best affiliate commission. I got tired of that. So I rented servers from Hetzner, DigitalOcean, OVHcloud, Vultr, Contabo, Hostinger, Kamatera, and a few others — installed the same tool on each one — ran the same benchmark harness — and wrote up what actually happened.

Sometimes the answer is what you’d expect (Hetzner is usually the best value). Sometimes it isn’t (OVHcloud Canada beats Hetzner US on RAG latency because it’s a shorter network hop to OpenAI’s east coast). Either way, the conclusion comes from the data, not from who sent me a nice email.

How we get paid

Some of the links on this site are affiliate links. If you click through one and sign up, the host pays me a commission — you pay the same price either way. That revenue keeps the lights on and pays for the servers I rent to run benchmarks.

The editorial calls are independent of the commission math. I rank providers by benchmark results and pillar pass/fail scores. Hosts that pay higher commissions do not win. Hosts that pay nothing (some of the ones I test don’t even have affiliate programs) win when the data supports it. The “skip this one” verdicts in our reviews are often on providers we make money from when readers sign up — we still say skip.

How the tests work

Every edition follows the same five-pillar framework: Setup, Performance, Pricing, Security, Support. Each pillar has 4-6 pass/fail criteria. We rent a fresh VPS from each provider (no reused credits, no sponsored instances), install the target tool via its official docker-compose or the provider’s one-click, and run a structured benchmark — sysbench for CPU, fio for disk IOPS, iperf3 for network, and tool-specific API calls for application-level latency.

All the benchmark scripts are open source at our methodology page. You can rerun them yourself with your own credentials.

Contact

Questions about methodology, corrections on a published number, or a request to benchmark a new tool: iam@arn.as.