To get to a position of the best optimised data centre that supports the business I work for, Quocirca Ltd, therefore advises that the following steps are taken:
1. Application rationalisation
Bringing the number of instances of applications under control is a great starting point for moving to a virtualised data centre. If you have different applications running or doing the same thing, choose a strategic vendor, and rationalise to its product wherever possible. If you have five different instances of SAP running, get it down to one or two.
The aim here is to take as much of the physical estate as possible and make it appear as a single pool of resources that can then be sliced up into logical units as required. This does not just apply to servers, but also to storage and the network.
However, certain workloads may still need to be placed on physical servers – either for distinct workload reasons or because the business is not yet ready to move to a completely virtual approach.
3. Application consolidation
Once you have the minimum number of applications and application instances running, it is time to consolidate them onto the smallest footprint possible. This can be carried out as a pre-virtualisation task, with the aim of minimising the number of physical servers being used to support each application.
In itself, this does have value, but Quocirca recommends that consolidation be carried out as a post-virtualisation task so that resources can be applied to workloads on a far more granular basis.
4. Application granularisation
The lack of functional granularity within the majority of enterprise applications is a brake on many organisations. Through the use of Web services and service-oriented architectures, you can surface the various functions within an existing application in a manner that allows each function to be called and used outside of the application itself.
This then enables business processes to be supported on the fly as functions are brought together to meet the needs of the process as it changes. New functions can be introduced either as new internal functions, or they can be sourced and implemented as external cloud services.
5. Data centre optimisation
Once an appropriate IT platform has been reached, this will impact the data centre itself. The business will have to review power distribution, cooling, space and other aspects to make sure that everything can still meet the needs of a far more densely populated environment. More targeted cooling will be required to minimise energy usage, along with floor-to-ceiling baffling to create suitable baffling for what should now be a far smaller overall data centre.
However, virtualisation and optimisation also provide the opportunity for the business to move towards far more ecological and long-term data centre efficiency approaches, such as free air or water cooling, energy reclamation and reuse and so on. The move can also be taken as an opportunity to review how functions are being provided to the business and whether the external cloud is a suitable means of providing new, or even existing, functions in a more appropriate manner to the business. The move to a new data centre architecture is the best time for such decisions to be made and to create a long-term, flexible architecture that will support the business for the future.
The aim of a highly virtualised data centre is to move away from a constraining, under-utilised platform to a far more dynamic, responsive and cost effective approach. New management tooling that can manage the mix of physical and virtual systems may well be required. However, the move to higher resource utilisations and densities, combined with lower energy costs and space requirements should help to move the IT budget from one in which 70% to 80% of a typical organisation's budget is spent "keeping the lights on" to one where less than 50% is spent in this way.
By freeing up 20% to 30% of the IT budget, the IT and facilities management (FM) departments can respond to the current "do more with less" calls from the business in a manner that still enables increased investment in what IT should be there for – to support the business in carrying out its core operations.Certain workloads may still need to be placed on physical servers – either for distinct workload reasons or because the business is not yet ready to move to a completely virtual approach.
Clive Longbottom, service director, Quocirca,
Sure, there is a distinct cost in making the move, but the business, along with IT and FM need to be aware of the cost of not carrying out such changes. The data centre increasingly becomes a brake on the business and business decisions cannot be made that reflect the needs of a dynamic market. Newer technologies and architectural approaches cannot be adopted in an effective manner, and the competition that does move in this direction becomes far more effective, attracting customers from your base and so lowering the overall effectiveness of the business in the market.
Neither will maintaining a mindset around applications sitting on physical servers help – as more applications are required, or as workloads increase, the end result will be the need to build a new data centre because of space, cooling or energy requirements. If the aim is to save money, suddenly having to find the money to build a new state-of-the-art facility would make retro-fitting an existing facility to meet the needs of virtualisation look like petty cash.
Make no mistake, virtualisation is happening – in many cases has already happened. As a cornerstone in ensuring that IT and FM departments support a business in the manner it demands, virtualisation has to be embraced, but with eyes fully open and with IT and FM working together in a strategic manner.
Clive Longbottom is a service director at UK analyst Quocirca Ltd. and a contributor to SearchVirtualDataCentre.co.UK.