Hosting Architectures: The Right Foundation for Every Application

Infrastructure is not a one-size-fits-all product. Depending on performance, isolation, and flexibility requirements, we deploy various hosting models – from dedicated hardware to virtualized private clouds. This page provides an overview of the underlying concepts and their suitable applications.

Dedicated Bare Metal Hosting

With bare-metal hosting, your application runs directly on physical hardware – without an intervening virtualization layer. This means maximum performance, as no computing power is “lost” to a hypervisor layer, and full control over the underlying resources. Bare metal is particularly suitable for applications with high, constant resource demands or special isolation requirements where no other instance ever uses the same hardware.

Public Cloud Hosting

Public cloud hosting is based on virtualized resources that can be flexibly scaled. We rely on OpenStack at OVH – an established, open-source cloud platform. The advantage: computing power, storage, and network can be adjusted as needed, new instances are available within minutes, and you only pay for what you actually use. This makes this model ideal for projects with fluctuating or growing resource requirements.

VMware vSphere Private Cloud Hosting

For organizations with higher demands for isolation and control, we offer private cloud hosting based on VMware vSphere. Here, your infrastructure runs on exclusively reserved hardware, virtualized via one of the most established enterprise virtualization platforms on the market. This combines the flexibility of virtual machines with the isolation of dedicated environments – and forms the basis for further compliance certifications (see our Datacenter page).

In addition to computing power, storage also determines how performant and reliable an application runs. We offer various storage concepts, depending on the chosen hosting model.

Block Storage on NVMe RAID

For our OpenStack environments, we use Block Storage with NVMe SSDs in a RAID 10 array. NVMe storage offers significantly higher read and write speeds compared to traditional SSDs, and RAID 10 additionally provides redundancy: if a drive fails, your data remains intact and operations continue without interruption. This combination is particularly suitable for applications with high data throughput, such as databases with many concurrent accesses.

Block Storage on Redundant SSD Storage

In the VMware private cloud environment, we use redundant SSD storage. Here too, data is mirrored or distributed across multiple physical drives, so a single hardware defect does not lead to data loss. This storage variant is designed for stability and reliability in an enterprise context.

S3 Object Storage

For files, backups, media content, or archive data, we also offer S3-compatible Object Storage. Unlike block storage, which functions like a classic hard drive, object storage is designed to store large amounts of unstructured data – images, documents, backups – cost-effectively and with virtually unlimited scalability. Its S3-compatible interface also allows for easy integration into many applications, for example, for automated backup strategies or media uploads.

Dedicated IP Block (Private Cloud)

For our private cloud environments, we have our own dedicated IP address block. This means we assign the IP addresses ourselves and can flexibly allocate them to individual virtual machines. If a VM fails or needs to be replaced, we can quickly redirect the address to another system if necessary – without changing the accessible IP for you. This ensures your service remains available under the same address, even if something changes in the background.

Floating IPs

In our public cloud environments (OpenStack), floating IPs are used: public IP addresses that can be flexibly assigned to a virtual instance and, if necessary, released and reassigned to another instance. This is practical when instances are exchanged, migrated, or temporarily moved for maintenance – without changing the publicly accessible address. This is the same principle as with the dedicated IP block, but based on OpenStack technology.

IPv6

We support IPv6 in parallel with classic IPv4, making us ready for the future. The available IPv4 address space is becoming increasingly scarce, while IPv6 is already standard in many network contexts – such as for international accessibility or in modern infrastructures – and is increasingly becoming a prerequisite.

The real strength of our approach lies in the flexible combination of hosting model, storage, and network concept – tailored to the respective application.

Dedicated Hosting (one VM = one application instance) is suitable when an application requires its own resources, needs to be scaled independently of other services, or has special isolation requirements – for example, for business-critical systems with high data volumes.

Shared Hosting, on the other hand, is the economical choice for applications with moderate resource requirements, such as classic WordPress or website hosting. Here, multiple websites share a common infrastructure base, which reduces costs without compromising the stability of individual websites.

Cluster and Load Balancer Architectures are used when an application requires high availability or needs to handle high load peaks. Instead of running on a single instance, traffic is distributed across multiple servers – if one fails, the others seamlessly take over. This is the right choice for applications where downtime would be particularly critical or where highly fluctuating user numbers are expected.

Which combination makes sense for you depends on your specific application, expected access numbers, and your requirements for failure safety – we would be happy to discuss this with you personally.