Classic managed hosting means a hoster sets up and runs a server and its software stack for you, and you ask the hoster when something on it should change. A cloud platform with an operations team gives your developers self-service access to deploy and change their workloads (through Kubernetes, CI/CD or GitOps), and puts engineers on call for the production running on top. How much of the infrastructure itself is self-service varies by provider. If you run a shop or a CMS on a known stack, managed hosting is usually the simpler and cheaper fit; if you ship a SaaS product with several services and deploy every week, the platform model fits better.
This guide compares the two models and German providers by model. Prices come up only to show the order of magnitude.
What is the difference between managed hosting and a cloud platform with operations?
| Classic managed hosting | Cloud platform with an operations team | |
|---|---|---|
| What you get | A server or server group, set up by the hoster with a stack it supports (typically PHP, a web server and a database) | An account on a cloud platform: Kubernetes clusters, object and block storage, load balancers, DNS |
| How changes happen | Through a control panel for common settings, often with SSH without root; anything beyond it is a request to the hoster | Your team deploys and changes workloads through Kubernetes, CI/CD or GitOps; depending on the provider, clusters and networking are self-service or set up by the provider; the operations team reviews and runs the platform work |
| Who runs what | The hoster runs the OS, updates and stack; you run the application inside it | The platform provider runs the hardware and platform services; the operations team runs clusters, monitoring, CI/CD and on-call; you run the code |
| Root access | Often not available, or limited | Full access to your own project and repositories |
| Scaling | Bigger plan or more resources, increasingly self-service in the hoster’s panel; mostly vertical, rarely automated | More replicas through the Kubernetes API; more nodes through the provider’s API or the operations team, automated on some platforms |
| Good fit | Shops, CMS sites, agency projects, one main application | SaaS products, several services, frequent deploys, containers |
A single Shopware shop on a classic stack gains little from running its own Kubernetes; a SaaS team that deploys ten times a day will find a request queue slow.
The line between the two is not sharp. Some managed hosters now run containers too: mittwald’s managed vServer can start your own Docker containers and has a public API, a CLI and a Terraform provider. A team with a few containers may not need either extreme.
When does classic managed hosting fit?
- You run an off-the-shelf application (Shopware, Magento, WordPress, TYPO3) and the hoster knows it well.
- Changes to the server are rare, and a control panel covers the common ones.
- You do not want to own containers, Kubernetes or CI/CD at all.
- Your traffic peaks are predictable and can be planned with the hoster.
German hosters here are mature: Profihost has run hosting for more than 20 years and tunes shop hosting for Shopware and Magento, mittwald builds for agencies, and Hetzner’s managed servers target web hosting, email and databases.
When does a cloud platform with an operations team fit?
- You build a product, not a website: several services, background workers, queues, object storage.
- Your developers want to deploy without filing a request each time.
- Customers or auditors ask for documented changes, access and incidents, and you have no platform team to produce them.
- Nobody on your team should carry the pager at night, but you still want to keep full access and the option to take operations back in-house.
Providers sell this as managed operations or a management level on top of a platform (plusserver calls its top level “Full Management”); enum calls it DevOps as a Service. It costs more per month than a managed server because it covers more: cluster upgrades, monitoring, pipelines and incident response. The comparison is with staff: a real 24/7 rotation needs several people (the Google SRE book puts the minimum for a single-site team at eight engineers), and German working-time law requires 11 hours of uninterrupted rest (§ 5 ArbZG), which night call-outs break. One or two in-house engineers cannot cover nights alone. For the full cost comparison against hiring, see managed Kubernetes vs. in-house operations.
How do German providers compare by model?
What each offer covers, from the providers’ own pages on 2026-10-09. Response times and support hours are only listed where the provider publishes them.
| Offer | Model | Who runs what | Out-of-hours support | Response time stated | Where it runs |
|---|---|---|---|---|---|
| Profihost Managed Server | Classic managed hosting | Profihost runs operation, updates and monitoring; you configure vhosts, databases and PHP in a panel | Support Mon to Fri, 08:00 to 18:00; 24/7 emergency support as an add-on | SLA PDF: response 2 to 6 hours within business hours, depending on the managed-service modules; 99.9% monthly availability with service credits | Group data center in Oberhausen, run by sister company KAMP, which holds the ISO 27001 certificate |
| Profihost Managed Solutions | Individual hosting setup with operations | Profihost runs setup, monitoring and documentation, with regular meetings | 24/7 reachability and emergency service as an option | ”Guaranteed response and intervention times”, details in the individual quote | Germany |
| Profihost Kubernetes ONE | Kubernetes platform | Profihost runs the platform underneath; you run clusters and workloads through a self-service dashboard | Production tier: 24/7 monitoring and incident response per SLA; Non-Production: best effort during business hours | No SLA figure on the page; a contact person promises answers “in under 2 hours” | German data centers, on OpenStack |
| IONOS hosting for WordPress | Classic managed hosting | IONOS handles technical and server-side administration of containerized WordPress hosting | 24/7 support | Not stated on the product page | Germany |
| IONOS Cloud (incl. Managed Kubernetes) | Self-service cloud platform | IONOS runs the platform; you run OS, clusters and applications. Support may not log in to your resources; OS and application support comes from partners | 24/7 support | Under 1 hour for a malfunction, under 72 hours for a service request | Germany, Spain, France, the UK and the US; Managed Kubernetes in the EU only in Frankfurt |
| Hetzner Managed Server | Classic managed hosting | Hetzner installs and runs Debian, Apache, MariaDB and PHP, with security updates and backups; no root access | 24/7 monitoring; phone support for managed servers Mon to Fri, 08:00 to 18:00 CET per Hetzner’s support page, plus email support | Not stated | Germany; Hetzner’s BSI C5:2020 Type 2 attestation lists Managed Server in scope |
| mittwald Managed vServer | Classic managed hosting for agencies | mittwald runs infrastructure, backups and software updates for managed apps; SSH access | Phone support Mon to Fri, 08:30 to 18:00; outside those hours a 24/7 emergency hotline for urgent technical problems | Not stated; 99.99% availability as an annual average, with a daily 23:00 to 06:00 maintenance window excluded | Own data center in Espelkamp, plus additional space in Frankfurt and Hannover |
| plusserver Managed Kubernetes with a management level | Kubernetes platform plus optional operations | You run clusters in self-service; plusserver sells management levels from Operational Support (24/7 monitoring, patching) to Full Management (24/7 administration); check whether the level covers your clusters | 24/7 German-language support stated | Not stated on the product pages | Own data centers in Germany; BSI C5 Type 2 |
| secunet cloud (SysEleven) MetaKube | Managed Kubernetes, self-service (MetaKube Core) or with operations (MetaKube Operator) | Core: you run clusters in self-service; Operator: secunet cloud (SysEleven) runs lifecycle management, monitoring and incident handling with 24/7 operation | 24/7 with MetaKube Operator | Not stated on the product page | Six data centers in Germany |
| enum DevOps as a Service | Cloud platform plus operations team | enum’s engineers run Kubernetes upgrades, node pools, monitoring, on-call, CI/CD, security patches, backups and restore tests; you keep the application code and the final say on production changes | On-call hours per contract | Down to 15 minutes for critical incidents, depending on the service level | enum in Frankfurt, one zone |
Checked October 2026.
Price models as of October 2026; check each provider’s pricing page for current prices. Profihost: server per month plus a Service & Support package; Kubernetes ONE per project tier plus worker units. Hetzner: monthly plus a one-time setup fee. mittwald: one-month minimum term. IONOS Cloud and plusserver Managed Kubernetes: platform usage. enum’s DevOps as a Service: monthly fee per contract (see DevOps as a Service), with platform usage on enum billed separately.
Managed servers start at tens of euros per month: Hetzner Managed Server from €34 plus a one-time setup fee of the same amount, Profihost Managed Server from €49 plus Service & Support from €49, mittwald vServer S at €43 (net, as listed on 2026-10-09). The platform-plus-operations model costs one to two orders of magnitude more, and it pays off only when it replaces engineering time.
Where others are stronger: Profihost, mittwald and Hetzner have run hosting for far longer than enum, which was founded in 2024, and they run classic PHP stacks with no container work on your side. Hetzner (BSI C5:2020 Type 2, scope includes Managed Server), plusserver (C5 Type 2) and IONOS Cloud (C5 Type 1, a point-in-time check of control design) hold attestations in their own name; enum holds no certification of its own today, its ISO 27001 is in progress with a target of Q4 2026, and C5 is planned for 2027. mittwald guarantees 99.99% availability for its managed vServer as an annual average, excluding a daily 23:00 to 06:00 maintenance window; enum does not publish an availability figure for DevOps as a Service on its pages. maxcluster, a Shopware and Magento specialist, publishes a 15-minute 24/7 response for critical incidents on its Enterprise service level (€350 per month on top of hosting) and 99.99% annual availability for network and server hardware. IONOS Cloud runs more regions, and secunet cloud runs six data centers. Profihost’s Kubernetes ONE and plusserver let you create clusters in self-service today; on enum, clusters are set up on request until self-service cluster creation ships in Q4 2026.
What should you check before you sign?
Ask every provider the same questions, and get the answers in the contract, not in a sales call.
- Who is on call, and when? “24/7 support” can mean a hotline that takes tickets, or an engineer who logs in and fixes the problem. Ask which one you get at 3 a.m. on a Sunday, and what it costs.
- Response time in the contract. For a production-down incident: measured from when, during which hours, and what happens when it is missed.
- What the operations team may touch. Some support teams may not log in to your systems at all; IONOS Cloud’s support agreement says so. Others run everything up to the application. Write down the line.
- How changes happen. Can your developers deploy and scale themselves, or does every change go through a request? Is there a record of who changed what?
- Exit and portability. What do you keep if you leave: configuration as code, runbooks, backups, repositories? A standard stack (upstream Kubernetes, S3-compatible storage) moves more easily than a hoster-specific panel setup. The EU Data Act sets rules for switching cloud providers; for managed operations, ask for the notice period, termination fees, handover duties and data export in writing.
- Where it runs, and who owns the company. Data center location, the legal entity you sign with, its parent (Profihost, for example, belongs to group.one), and whether certificates belong to the provider or the building.
- Scope. Is the database included? Backups of application data or only the server? Restore tests?
- Data processing and regulatory points. Ask for the GDPR Art. 28 data processing agreement and the list of sub-processors. If NIS2 or DORA applies to you, have your counsel check the contract against them.
If you are leaving a host because of outages, ask for its incident history for the last 12 months. Our guide to reliable managed Kubernetes in Europe covers SLAs and status pages.
This guide is general information, not legal advice. Have your counsel assess your specific situation.
Is DevOps as a Service an alternative to IONOS or Profihost?
It depends on what you run. For a single shop or CMS on a classic stack, a managed hoster is usually the better and cheaper fit. For a product team that has outgrown a managed server, or runs Kubernetes where nobody operates the layer above the control plane, DevOps as a Service fills that gap.
Limits with enum:
- Its operations team runs platforms on enum only, with the migration part of onboarding (see how migration works).
- Applications need to run in containers; if yours do not yet, agree in the assessment who prepares them and how.
- Production runs in one Frankfurt site with a single zone (fra-a). Redundancy is across failure domains inside that site, not across availability zones.
- enum has no managed databases, caches or queues, so agree whether your database runs in the cluster under the operations team or elsewhere. enum also has no volume snapshot or backup product; backups are part of the operations work.
- The published platform SLA is 99.9% for the Kubernetes control plane, with no worker-node SLA and no published credit terms.
- Clusters are set up on request until self-service cluster creation ships in Q4 2026, and there is no cluster autoscaling.
- Support included with every account covers working days. Around-the-clock response for critical incidents comes from Enterprise Support, or from a DevOps as a Service contract whose on-call hours include nights and weekends.
- enum has no customer-facing audit log or fine-grained roles yet (both planned for Q4 2026), and no Terraform provider yet (planned for Q4 2026). Until then, keep the record of who changed what in your Git history and CI/CD logs.
- enum has a GDPR data processing agreement in every contract but publishes no sub-processor list.
Where enum fits
enum is a European public cloud platform run by enum GmbH, with one production region in Frankfurt. DevOps as a Service puts enum’s engineers on your platform (upgrades, monitoring and on-call, CI/CD, security updates, incident response), with response times down to 15 minutes for critical incidents depending on the service level you choose. For business applications an IT department wants hosted and operated, see the Managed Hosting page. You keep full access to your enum project and repositories. enum GmbH is a German company with no US parent and no US company in its structure. enum’s own ISO 27001 certification is in progress, with a target of Q4 2026; the Frankfurt data center’s ISO 27001 and EN 50600 certificates belong to the facility operator. If you are comparing against IONOS specifically, see enum as an alternative to IONOS.
FAQ
What is the difference between managed hosting and DevOps as a Service? Managed hosting runs a server and its stack for you, and you request changes. DevOps as a Service runs a cloud platform (usually Kubernetes) for you, while your developers keep self-service access to deploy and change their workloads.
Is managed hosting enough for an online shop? Usually yes. A Shopware, Magento or WordPress shop on a classic stack is what these hosters build for. Check out-of-hours support before peak season; some offers sell it as an add-on.
Does “24/7 support” mean someone fixes my system at night? Not necessarily. IONOS Cloud’s support, for example, answers malfunctions within an hour but may not log in to customer resources, and some hosters offer only an emergency hotline outside business hours. Ask whether an engineer works on your system.
Can I get managed Kubernetes with 24/7 operations in Germany? Yes. plusserver sells managed Kubernetes together with management levels up to Full Management with 24/7 administration; secunet cloud (SysEleven) offers MetaKube Operator, with 24/7 operation and lifecycle management by secunet cloud; Profihost’s Kubernetes ONE has a production tier with 24/7 monitoring and incident response per SLA; and enum’s DevOps as a Service runs Kubernetes on enum in Frankfurt with on-call hours and response times agreed per contract, without a published standard 24/7 commitment. Compare what “24/7” covers in each contract.
What should a startup without a platform team choose? If you ship a product with several services, a cloud platform with an operations team usually fits better than a managed server, because your developers can deploy themselves. Compare the monthly fee with the cost of one engineer, and check whether you qualify for enum for Startups credits.
Sources
Checked 2026-10-09: Profihost, Profihost Managed Server, Profihost SLA, KAMP ISO 27001 certificate, Profihost Managed Solutions, Profihost Kubernetes ONE, IONOS hosting for WordPress, IONOS Cloud support agreement, IONOS Managed Kubernetes locations, IONOS Cloud certificates, Hetzner Managed Server, Hetzner support hours, Hetzner BSI C5 summary, mittwald Managed vServer, mittwald customer service hours, mittwald terms, plusserver Managed Kubernetes, plusserver management levels, plusserver certificates, secunet cloud, secunet cloud MetaKube, maxcluster SLAs, enum DevOps as a Service, enum Managed Hosting.