Contabo: enormous specifications, contested reality

Eight gigabytes and four cores for six euros. The RAM is real, the disk throughput is the argument, and whether it suits you depends entirely on your workload.

Contabo occupies a strange position in the hosting conversation. Its specification-per-pound is the best on the open market by a wide margin — 8 GB of RAM and several cores for around €5.50 — and its reputation among people who benchmark things is mixed at best. Both of these are fair.

What is genuinely true

The company is large and long-established. Munich-based, trading since 2003, with facilities in Germany, the US, the UK, Singapore and elsewhere. This is not a one-person operation that vanishes at Christmas — whatever else, it is not the low-end market's disappearance risk.

The RAM is real. You get what is advertised, and for memory-hungry workloads that is the whole story. A development box, a Kubernetes lab, a Minecraft server, a Nextcloud instance, an Elasticsearch node for testing — anything that wants a lot of memory and does not care about I/O latency is excellent value here.

There is a London location. Genuinely useful, and rarer at this price than people expect.

What the criticism is about

Disk and CPU contention. The consistent complaint, across years and across node generations, is that storage throughput and CPU steal vary enormously depending on which host machine you land on. Two identical plans can benchmark very differently. For a database-driven site — which is what WordPress is — inconsistent I/O is precisely the thing that produces mysteriously variable page load times.

The setup fee. Most plans carry a one-off charge, which makes short-term use worse value than the monthly figure suggests and quietly raises the real cost of trying it.

Support pace. Slow by European standards. Not absent, but not the same experience as netcup or LiteServer at similar money.

Backups and snapshots cost extra, and are off by default.

How to judge it for yourself

Order monthly, then benchmark on day one before you commit anything real to it:

# random read/write IOPS — the number that matters for a database
fio --name=rand --ioengine=libaio --rw=randrw --bs=4k \
    --numjobs=4 --size=1G --runtime=60 --group_reporting

# CPU steal: watch the 'st' column under load
vmstat 1 20

# sequential throughput, as a rough sanity check
dd if=/dev/zero of=test bs=1M count=2048 conv=fdatasync

If steal time is consistently above a few percent, or random I/O is poor, you are on a busy node. Rebuild in a different location or ask to be moved — and if the second attempt is no better, take the answer.

The comparison that matters

netcup is the direct comparison most people miss — similar specifications, better consistency, similar price, no UK location. If you do not need London, it is usually the better buy.

The verdict

Buy Contabo for memory-heavy, latency-tolerant workloads where the specification genuinely is the point, and where a variable disk will not ruin your day. Labs, build servers, media, storage, self-hosted services.

Do not buy it for a production WordPress site you care about, particularly a shop. Inconsistent I/O is exactly the wrong failure mode for a database-driven site, and £5 saved is not worth a checkout that intermittently takes four seconds. Cloudways or a decent shared host is the better answer there.

Contabo is not overselling in a way it hides — the plans are what they say. It is selling a different trade-off from Hetzner: more resources, less consistency. Know which of those your workload actually needs.

Published 28 March 2026. Prices and ownership change — check the comparison table for current figures, and the provider's own site before buying. No affiliate links anywhere on this site.