Review: 6 months with Contabo — boring in the best way

Tl;dr: it just works. That's the highest praise I can give a budget host, IMO.

Specs: KVM 2 vCPU / 4 GB / 80 GB NVMe / 2 TB @ 1 Gbit, $7.20/month, FRA location.

Uptime: 182 days, one planned migration (30 min notice, finished in 8 mins). The graphs don't lie — 100% excluding that window.

Tickets:

• Month 2: asked about rDNS setup, answered in 47 minutes, resolved same day.
• Month 4: thought I had a routing issue, turned out I misconfigured my netplan gateway. They still checked their side before pointing me to my typo. No attitude, just "happens more than you'd think."

Peak IO: fio randread 1.2 GB/s, randwrite 380 MB/s. Not amazing, not bad. YMMV during busy hours, I only run low-traffic stuff.

Cons: no BGP, no custom ISO upload yet. IPv6 is /128 only, no /64. Billing portal is functional but ugly.

Would I renew? Already did.

Hi quinnbizz — thank you for the detailed review and for being a customer these past six months. Really appreciate the honest, balanced take.

A few quick notes from my side:

• RDNS: they recently added self-service in the panel, so no more tickets needed for that.
• IPv6: /64 delegation is on the roadmap for Q4, no firm date yet but it's actively being tested.
• Custom ISO: you're right, not available today. They have a netboot.xyz-based rescue environment that covers most cases, but full ISO upload is planned for early 2027.

Re: that netplan misconfiguration — their support team has a soft spot for "it's always DNS (or gateways)." I see it come up weekly in the tickets I read.

If anyone wants specific benchmarks from a particular node or has questions about routing, feel free to DM me or drop a note here. I'll check what I can share from public sources.

— Jane

52 boxes across 9 providers and I keep coming back to "boring is good"

Contabo's on my shortlist but that €7.20 for 4GB puts them at $1.80/GB which is... fine? Not amazing. Compare to my OVHcloud black friday special, €4.50 for 3GB = €1.50/GB. Or Vultr's recurring $6 for 4GB = $1.50/GB.

But OVHcloud's network is 500 Mbit not 1 Gbit and their panel breaks monthly. Vultr's support takes 2 days. So boring wins I guess.

Peak IO question though —

Also jane any plans for storage VPS? I'd throw $3/month at a 400GB rust box for backups in a heartbeat. Don't need NVMe for that.

Respect.

I'm 14 months with Contabo on their 1 vCPU / 2 GB plan. 99.997% measured from three external monitors (HetrixTools, UptimeRobot, my own smokeping). The 0.003% was a 90-second blip during that migration quinnbizz mentioned — I got the same 30 min notice.

Alert fatigue is real with budget hosts. I used to monitor 23 providers and paged myself into insomnia. Pruned to 6 now. Contabo's one of two where I disabled SMS alerts entirely. Status page is basic but updated honestly, unlike some who mark "degraded performance" as "all systems operational" for 6 hours.

Re: peak IO — I don't run heavy disk workloads but I do see CPU steal 3-8% during 20:00-23:00 CET. Not enough to hurt my stuff, but it's there. Would be curious if others see worse.

My only real gripe: no API for firewall rules. I automate everything and clicking a panel for

Would the 2GB plan work for hosting a small nextcloud for like 3 people? I don't need much storage, just photos and some documents. My budget is basically $5/month max

Also jane do you have any student discounts or anything? No worries if not.

I've been on oracle free tier for a year but they're so annoying with the always-free limits and I keep getting emails about "potential abuse" when I'm literally just... using it. Tired of checking if my instance got terminated every morning.

Thanks everyone for all the info, this thread is super helpful!

Route via cogent in FRA adds 12ms versus their level3 path, but both are stable. My traceroute from stockholm:

traceroute to contabo.com (203.0.113.47), 30 hops max
 1  gw.local (192.168.1.1)  0.5 ms
 2  isp-core (198.51.100.1)  4.2 ms
 3  level3-peer (203.0.113.1)  11.3 ms
 4  contabo.com (203.0.113.47)  23.7 ms

Compared to Hetzner's FRA at 31ms via tata. Contabo wins on latency, loses on nothing measurable from here.

Re: sarahjey — 2GB is tight for nextcloud with default settings. You'll want to tune mysql/mariadb memory limits, maybe use sqlite if truly tiny. Or just pay the extra €2.50 for 4GB and not fight it.

1. Resource estimation for Nextcloud: base memory footprint is approximately 512 MB with PHP-FPM, 256-384 MB for MariaDB with default buffers, 128 MB for Redis (recommended for file locking). 2 GB is technically viable for 3 users with aggressive tuning, but leaves no headroom for OS processes or spikes.

2. Oracle free tier termination risk: their A1 instances are subject to capacity reclamation with 30-day notice, but many users report unannounced terminations. This is documented behavior, not a bug.

3. Contabo specifics from this thread:

• CPU steal 3-8% during peak (uma's data) is acceptable for non-burstable plans at this price point. AWS t3.small averages 5-15% in my experience.
• NVMe performance 1.2 GB/s read, 380 MB/s write suggests Phison E12 or similar entry-level controller, not datacenter-grade but adequate for stated workloads.
• RDNS self-service and pla

What CPU model are they actually running on those 2 vCPU plans

Boring really is the best praise for budget hosting

Anyway I checked my Contabo Mumbai box: AMD EPYC 7282. The Nuremberg one I had before was Intel Xeon E5-2630 v4. So it varies by location, not just plan.