What is Virtualisation?

Traditionally infrastructure architects have been faced with a tough choice when designing server infrastructure to host applications:

• Multiple applications on a single operating system (OS) on a single server.

With this configuration we would expect a much greater utilisation of the server hardware, because multiple applications are sharing the resources (CPU, SAN, network, RAM) on the server. Maximising the utilisation of the hardware spend is increasingly becoming a major concern as even entry and mid level servers are loaded with more processing power that ever before, so this seems like a good solution.

The core issue with this approach comes down to maintenance. If one application requires an upgrade to new patch level, or a new release of the .Net framework, or an operating system upgrade, will the other applications be compatible with this new system configuration? Will all the other applications require full testing to ensure there are no knock-on effects? Who pays for the upgrade testing - the owner of the application being upgraded, or all the application owners sharing the server even if they did not request the change? If the upgrade cannot be applied to allow the testing to happen, will that leave possible security holes or stability issues for the unpatched application/operating system? 

Complicating this approach further are the agreed outage windows. Ideally the applications would all have the same approved outage windows and the same hours of operation (e.g. 7-7, 24x7), but this is not always the case. If an application requires a full weekend outage, will the business owners of the other applications agree? Shifting applications to another physical server during the upgrade is not a trivial exercise, and in itself will normally require an outage and significant planning and change management.

• One application on a single operating system (OS) on a single server.

With a single application on a single server there are no application interdependencies or maintenance conflicts to contend with. Upgrades, version changes, even hardware maintenance cah be scheduled with the agreed outages windows, and testing need only concern the application being upgraded. 

Our observation is that most one application one server configurations consume between 5 and 10 % of that server's processing power during normal load.  The rest is unutilised and unutilisable. Of more concern is that datacentre managers find their datacentres rapidly filling up with new servers everytime a new application or system is deployed.

This approach also has implications in terms of managing datacentre space, cooling capacity, electricity consumption, the provision of backup power generation, the space required to run cabling and the management of all the network connectivity. This approach therefore conflicts with the trend towards "green" practices and the minimisation of impact on the environment.

Our Approach - Virtualisation

In technical terms virtualisation is the abstraction of the operating system from the physical hardware it runs on. In everyday terms, virtualisation allows multiple operating systems, each of which are completely separated and isolated, to run on the same physical hardware and share those hardware resources (CPU, RAM, disk, network). These virtualised operating systems, or "virtual machines", "believe" that they are the only machine running on that physical server.

Virtualisation means that we CAN put multiple applications, each within their own operating system instance, on the same physical server. Each application is isolated within its own operating system, completely separated from the other applications, which reside in their own individual operating systems.

The upgrade of an application, or the operating system on which it resides, can be carried out without affecting the other applications and operating systems. The application and its operating system can be rebooted, reconfigured, allocated additional resources and upgraded without an outage or testing required for the other applications.

Should the physical server require an outage for a hardware or firmware upgrade, or for a datacentre move, virtualisation technologies (e.g. VMware's ESX 3.0 VMotion) enable the dynamic realtime movement of applications and their operating systems from one virtualised server to another virtualised server, with almost no outage required. When the upgrade is completed, the application can be moved back to its original location.

Because hardware utilisation can be maximised, the requirements for datacentre space, cooling, power, power protection and infrastructure are greatly reduced. 

Developing a virtualisation strategy to migrate your physical environment to a virtual environment is always the first and most essential step.  RHE IS provides expert technical and business advice to assist our customers to ensure their virtualisation strategy will address their business needs and requirements. The RHE IS Virtualisation Approach phases are:

• Identify the business and technical goals of the Virtualisation strategy.
• Carry out a virtualisation assessment and develop the target virtualisation architecture.
• Identify an initial application or system within the organisation where a high level of business benefit and return on investment can be demonstrated with virtualisation.
• Progressively build the virtual infrastructure and migrate the physical servers to virtual servers in an ordered and staged approach to manage risk and business impact. 
  

Key Benefits to our Customers

• Lower your number of physical servers – but utilising the full processing potential of a server you can run multiple operating systems on a single server.
  
• By implementing a server consolidation strategy with virtualisation, you can increase the space utilisation efficiency in your data centre.
  
• By having each application within its own "virtual server" you can prevent one application from impacting another application when upgrades or changes are made.
  You can develop a standard virtual server build that can be easily duplicated and deployed
  
• You can deploy multiple operating system technologies on a single hardware platform (e.g. Windows Server 2003, Linux, Solaris x86).