Transparent benchmark with no result upload

Linux Server Benchmark: Open-Source CPU, Disk and Network Test

A single score is rarely enough when comparing servers. Eka Sunucu Benchmark exposes CPU, memory, storage and network behavior separately. System information and a short CPU measurement run by default; disk and network tests are opt-in.

MITBashfioiperf3OpenSSL

root@server:~$

curl -fsSL https://www.ekasunucu.com/templates/website/EkaSunucu/assets/eka-sunucu-benchmark.sh -o /tmp/eka-sunucu-benchmark.sh && bash /tmp/eka-sunucu-benchmark.sh --hizli

Build test command

Test profile

iperf3 target serveriperf3 port

Command to run

curl -fsSL https://www.ekasunucu.com/templates/website/EkaSunucu/assets/eka-sunucu-benchmark.sh -o /tmp/eka-sunucu-benchmark.sh && bash /tmp/eka-sunucu-benchmark.sh --hizli

Open script source

Download and inspect first

Before running a shell script downloaded from the Internet, inspect its source. The command below downloads the script and prints it in the terminal without executing it.

curl -fsSL https://www.ekasunucu.com/templates/website/EkaSunucu/assets/eka-sunucu-benchmark.sh -o eka-sunucu-benchmark.sh && sed -n '1,240p' eka-sunucu-benchmark.sh

✓

Security behavior

The default mode does not open a benchmark connection to an external host and does not upload results. Disk testing creates a temporary 512 MiB file only with --disk/--tam and removes it afterward. Network testing connects only to the iperf3 target you provide.

No root requirement
No automatic package installation
No result upload
No default benchmark network connection
Temporary disk file is removed automatically
The user chooses the test target

What does it measure?

SYS

System information

Distribution, kernel, architecture, hostname, virtualization, CPU model/topology, RAM, swap, root filesystem and block devices.

CPU

Short CPU measurement

Uses OpenSSL SHA-256 speed as a short repeatable CPU micro measurement. It is not an application benchmark, but it is a useful signal when comparing similar systems under comparable conditions.

I/O

fio disk test

When fio is available, the script runs 4K random read/write and 1 MiB sequential read/write tests so IOPS and throughput can be viewed separately.

NET

iperf3 network test

Network testing never starts automatically. With --ag or --tam and a target iperf3 server, it runs an 8-second client test and avoids silently connecting to arbitrary public test endpoints.

How to read the results

CPU: Run the same command at least 2-3 times under comparable load; do not treat one run as a final verdict.
Disk: 4K random IOPS and 1M sequential MB/s represent different workloads. A database and a large-file transfer should not be judged by the same metric.
Network: Port capacity is not the same as throughput to one remote endpoint. Target location and network path change the result.
VPS: Neighboring VM load and hypervisor policies can change results at different times.

Common benchmark mistakes

Comparing different benchmark sizes directly
Treating cache effects as real disk performance
Generalizing one benchmark score to all application performance
Testing one server under load and the other while idle
Ignoring that the iperf target can itself be a bottleneck
Ignoring RAID, filesystem and virtualization layers

Official technical sources

fio DocumentationOfficial Flexible I/O Tester documentation ESnet iPerf3Active IP network performance measurement OpenSSL Documentationopenssl speed and cryptographic tools

Frequently asked questions

Does the benchmark script require root?

No. Most system information is readable as a normal user. fio only requires write permission to the selected temporary directory.

Does the script install packages or change system settings?

No. This version performs no automatic package installation and makes no persistent system setting changes. Missing fio or iperf3 causes the related test to be skipped with a message.

Should I run the disk test on a production server?

I/O benchmarks can affect production performance. Run the disk profile during a low-traffic window and only when sufficient free space is available.

Why is iperf3 not enabled by default?

A network benchmark should make its destination explicit. The script therefore starts no network test until you provide an iperf3 target.

Compare your server with Eka Sunucu

Run the same command on two servers and compare CPU, disk and network in a similar time window and load state.

VPS servers Contact

Custom Server Services

Standard Servers