Our infrastructure, in real numbers
Every figure on this page was measured on our production server — the same platform VPS customers get — with standard commands you can run on any Linux box. No marketing averages, no "up to". One measurement, one date, one command.
Network latency from the data center to the world
TCP handshake time to port 443 (median of 5) from our server in Israel to major endpoints. The first two rows are edges that sit inside Israel; the rest show what an Israeli visitor pays when their server lives abroad.
| Target | Location | Latency | |
|---|---|---|---|
| 1.Cloudflare (1.1.1.1) | Israel edge | 1.64 ms | |
| 2.Google DNS (8.8.8.8) | Israel edge | 2.91 ms | |
| 3.AWS eu-central-1 | Frankfurt, Germany | 106.66 ms | |
| 4.AWS eu-west-1 | Dublin, Ireland | 119.96 ms | |
| 5.AWS us-east-1 | Virginia, USA | 195.94 ms | |
Disk and CPU
Reads above 5 GB/s mean heavy MySQL index scans and cold-cache WordPress loads never wait on disk. 1.5 GB/s synced writes mean backups, logs and WooCommerce orders do not queue.
The platform that was measured
| CPU (host node) | Intel Xeon Silver 4210R @ 2.40GHz |
| Virtualization | Xen, dedicated resource allocation |
| Operating system | Debian 13.6, kernel 6.12.73+deb13-amd64 |
| Measurement server | 2 vCPU, 11 GB RAM — the production server serving this site |
| Last boot | 06/03/2026 |
Methodology — run this at your provider
These are the exact commands. They run on any Linux, need no installation, and are safe on a live server (512MB, two CPU seconds, five TCP connects). Compare your output to the numbers here.
# Disk (512MB, direct I/O)
dd if=/dev/zero of=/tmp/t bs=1M count=512 conv=fdatasync
dd if=/tmp/t of=/dev/null bs=1M iflag=direct && rm /tmp/t
# CPU (single core, AES-256-GCM)
openssl speed -seconds 2 -evp aes-256-gcm
# Network (TCP handshake to :443, run 5x, take the median)
curl -4 -s -o /dev/null -w "%{time_connect}\n" https://1.1.1.1/
curl -4 -s -o /dev/null -w "%{time_connect}\n" https://s3.eu-central-1.amazonaws.com/
# Uptime
uptime -sMeasurements run on an active production server under real load — not an empty demo box. The date at the top is the latest measurement; data refreshes on every site deploy.
Questions about the measurements
Why measure on the production server and not a clean one?
Because that is what a customer actually gets. An empty demo server always looks better. The server measured here serves linux-hosting.co.il live, with Node, nginx and backups running in the background.
Will my VPS get exactly these numbers?
Same platform, same nodes, same storage. Disk and network will match; CPU throughput scales with the vCPU count you choose. These figures are from a 2 vCPU server.
Why TCP and not ping?
Large cloud providers filter ICMP intermittently, so ping shows 100% packet loss even when the network is fine. TCP handshake time is what a browser actually experiences, and one tool works against every target.
Want these numbers for your site?
Build a VPS on the platform measured here, or send us the output from your current server and we will tell you honestly whether there is anything to gain.
