Benchmark comparison: YABS on 15 providers, same $3.50 price point

I ran YABS on every VPS I could find at exactly $3.50/mo or equivalent annual. Same region (EU-West), same time of day, same Debian 12 image. The graphs don't lie.

Raw data:

Provider        CPU    Mem    Disk    SeqRead  SeqWrite  4kRandR  4kRandW  GB5
OVHcloud       2      2G     40G     412      380       12.4k    8.2k     1423
Vultr        2      2G     50G     198      176       4.1k     2.8k     891
Hetzner      1      1G     20G     156      144       3.2k     1.9k     612
InterServer         2      2G     40G     398      371       11.8k    7.9k     1389
Hostinger      2      4G     60G     205      189       4.5k     3.1k     943
KnownHost      2      2G     40G     387      362       11.2k    7.6k     1356
RackNerd       1      1G     25G     167      152       3.5k     2.1k     634
GreenCloudVPS        2      2G     40G     401      376       11.9k    8.1k     1401
HostHatch        2      2G     40G     394      368       11.6k    7.8k     1376
CloudCone      2      2G     40G     396      374       11.7k    7.9k     1384
Leaseweb        1      2G     30G     178      161       3.8k     2.4k     678
Time4VPS        2      2G     40G     389      363       11.3k    7.7k     1362
Contabo       2      2G     40G     395      372       11.5k    7.8k     1378
Vultr-P      2      2G     50G     201      179       4.2k     2.9k     902
Hetzner-P    1      1G     20G     159      147       3.3k     2.0k     621

Two clear clusters. The top cluster is suspiciously tight. I think some of these are temporarily upgrading test nodes. Need to verify.

Those top numbers look like cached or NVMe-over-provisioned test instances.

• OVHcloud: 412 MB/s seq read? That's datacenter SSD, not budget VPS
• InterServer, GreenCloudVPS, HostHatch, CloudCone, Time4VPS, Contabo: all within 5% of each other
• Same CPU GB5 too? 1356-1423 range is identical silicon

Either same upstream (possible) or same bait-and-switch (probable).

This methodology raises GDPR concerns if you did not obtain explicit consent from providers before automated testing. Article 5(1)(a) requires lawful basis for processing. Did you consider legitimate interest assessment?

Also your timestamp collection may violate provider logging policies. EU Digital Services Act Article 13 obligations apply to hosting providers. Your benchmark script may constitute unauthorized access under national implementations.

15 providers, same price point, same region. You're hitting shared infrastructure during off-peak. Numbers will vary 20% hour to hour.

No DDoS angle here. Carry on.

Please explain, server going to lie about spec? Going to cheat on benchmark? Very bad honor.

Hana5 going to test with real workload, not fake number