Skip to content

My 'unlimited' bandwidth got limited

Reviews by center807 6 replies 549 views
5 #1

Our enterprise-grade network operates on N+1 redundant feeds with enterprise-grade capacity planning. All enterprise-grade accounts include enterprise-grade bandwidth subject to our enterprise-grade acceptable use framework.

Your account triggered enterprise-grade automated review when sustained throughput exceeded enterprise-grade baselines for your enterprise-grade tier. This is standard enterprise-grade practice across enterprise-grade hosting.

The TOS references "reasonable use" to ensure enterprise-grade performance for all enterprise-grade customers on our enterprise-grade platform. Enterprise-grade mitigation was applied proportionally.

For dedicated enterprise-grade transit pricing, sales can quote enterprise-grade commits. Our enterprise-grade team is available.

Location and specific capacity figures are enterprise-grade confident

3 #2

Sustained 94mbps for 18 days; 14.7TB total; ticket open; no response for 72 hours; throttled to 10mbps at 2:47 UTC; graphs show hard drop; not gradual; they lied

#3
Speedtest by Ookla

Server: RackNerd - Amsterdam (id = 14293)
ISP: Contabo
Idle Latency:    12.34 ms   (jitter: 1.2ms, low: 11.1ms, high: 15.6ms)
Download:        9.87 Mbps (data used: 12.4 MB)
Upload:          9.91 Mbps (data used: 11.8 MB)
Packet Loss:     0.0%

Server: HostHatch - Amsterdam (id = 28901)
ISP: Contabo
Download:        9.92 Mbps
Upload:          9.85 Mbps

Decent disk IO, meh network. Pre-throttle I got 940/940 on same test.

Comparison:

  • Contabo "unlimited": 14.7TB then hard cap
  • Hetzner 5TB plan: actually 5TB then €0.01/GB
  • Hostinger "fair use": undefined, but I did 22TB once with no issue
  • GreenCloudVPS: publishes actual numbers in TOS

fio, iperf, geekbench. results or gtfo.
#4

This is why I only recommend providers that specify virtualization tax explicitly.

Contabo uses KVM with virtio-net, which has near-zero virtualization tax for networking. The throttle isn't a virtualization issue - it's policy. They applied traffic shaping via tc or Open vSwitch, not a resource constraint.

"Reasonable use" without numbers is worse than OpenVZ. At least OpenVZ tells you your beancounter limits. This is opaque policy enforcement with no cgroup visibility.

Check if they used HTB or policing. You can detect shaping with `iperf3 -R` and looking for consistent flat-top throughput. Real congestion shows variable rates.

Virtualization tax on CPU was probably minimal here. The tax is on transparency.

virsh list --all | wc -l: 47
#5

10mbps is roughly 3.3TB/month. They didn't nullroute, they shaped. That's scrubbing-center thinking applied to billing.

Attack comparison: 14.7TB in 18 days = 76 Gbps average sustained. A real ddos at that size gets filtered at 95th percentile, not hard-capped. Anycast would spread it. Scrubbing would clean it.

They chose cheapest option: tc qdisc. No anycast, no scrubbing, just rate limit. Not even burst allowance.

Fiber providers do this. Transit costs are real. But "unlimited" with invisible cap is false advertising, not mitigation.

mitigated 800Gbps before breakfast
#6

Guys this is why I stopped pushing Contabo to my clients. Margins were good but support tickets eat all profit. Any leads on provider with actual unlimited and decent panel? I tried KnownHost but thier API is broken, recieved 500 errors for 3 days. Hostinger no API at all. Need something for small business guys who actually use bandwidth.

your margin is my opportunity
#7

Same issue with Contabo, 11TB then cut

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft