A fixed location for dedicated cloud servers

vom 16.10.2019

16.10.2019 I by Henrik Hasenkamp / Elke Witmer-Goßner

Cloud infrastructure is flexible and dynamic. However, compliance rules in some industries require dedicated servers that have been at odds with the cloud. This is now changing with special solutions for virtual locations.

A high-performance server that is available to a company around the clock every day of the year? In view of modern, virtual cloud infrastructures, this seems like a thing of the past. Dedicated servers have a disadvantage that seemed to be overcome by the cloud: processor cores spend long hours of the day in idle mode with such a device. In most IT organizations, this is now considered a waste of resources.

Nevertheless, dedicated servers still have their authorization today. Many companies want to use them for individual workloads in order to have a certain amount of resources available at all times - for example, 64 cores permanently, at least 128 GB RAM or no less than eight terabytes of disk space. This demand exists in many industries that process sensitive data, such as healthcare, insurance, and finance. Here, however, dedicated IT resources are necessary for compliance reasons alone. In addition, there are many time-critical tasks that must be processed with high priority.

For example, insurance companies perform certain tasks when processing their contract databases in a precisely defined time window. Here every minute counts, even short waiting times for the allocation of new resources burst the company's schedules with frequent "scaling jumps". This is why such organizations rely on IT resources that are only available to them and fit better into their existing processes. This requirement is met in hosting data centers by dedicated hardware. Dedicated server users actually access a specific hardware server. It provides the desired performance data and is only available to this customer as a fixed network resource. The single machine is not used by anyone else and works 24×7 even without workloads. From the user's point of view, it looks as if you have the corresponding hardware in your own server room.

Dedicated servers can also exist from the cloud

The opposite of this is the modern, virtual IT infrastructure in the cloud. "Infrastructure as a Service (IaaS) offers virtual cloud services that are dynamic and flexibly scalable - in all directions. The servers, the number of cores used, the available RAM and the storage capacities used are basically metaphors. They stand for an infrastructure that is software-defined and put together according to use. On the hardware level, it can happen that processor cores, RAM and storage are distributed on completely different servers.

There is a clear gap between these two IT deployment options, especially with regard to the service level, because the dedicated server can fail. This is rare because the hardware is redundant - for example, a defective hard disk can be removed during operation without interrupting the workload. Nevertheless, with dedicated hardware it is difficult to guarantee an availability of one hundred percent.

The best values are achieved by high-availability data centers. They guarantee an availability of 99.9999 percent. This is availability class 6, which corresponds to about 31.6 seconds of downtime per year. But a lot can happen in half a minute. If, for example, a database workload restructures tables at the time of the malfunction, the entire database may be defective. Now when the server becomes available again, backups must first be installed - a further delay that can add up to hours with other work on software and data.

Reservation ensures availability

gridscale's infrastructure solution, for example, which offers 100 percent availability, avoids these difficulties. This is possible because the virtual infrastructure is distributed over several data centers and sufficient hardware resources. If there is a failure, the user will not notice anything. The management layer of the cloud solution shifts the corresponding resources in real time to other hardware elements. It is precisely this high reliability that is one of the most important reasons for using virtualized cloud infrastructures.

But based on demand, the question arises: Can the advantages of a flexible cloud and exclusive hardware server also be combined without taking over the disadvantages of a lower service level? The answer is virtual locations. They offer a direct link between the server booked by the user and the underlying hardware resources. A correspondingly large physical device is assigned to the virtual device. Put simply, the operator reserves one of its servers in the data center for a single customer.

There is an important difference to virtual servers: a new dedicated server is only available with a short delay because the corresponding resources must first be freed and then reserved for the user. In every other respect, the server behaves as usual, but it has a specific "location" in the cloud service data center. gridscale has therefore christened its forthcoming service "Virtual Locations". It can be booked under the infrastructure provider's simple user interface parallel to traditional virtual servers and works seamlessly with them. It is available on the familiar user interface with the same automation features as a virtual server. Only when booking is it not possible to bill to the minute, only to the day.

* The author Henrik Hasenkamp is CEO of gridscale GmbH.

The original article in german can be found here.

    Back to overview