The virtualised data centre gives us an opportunity to review where we are in terms of using IT as a platform that supports the business. Sure, virtualisation can be used to drive up resource utilisation while lowering energy requirements -- but if that is all it is used for, are organisations missing out on some hidden opportunities?
Any resource calculation is aimed at ensuring that the application can meet at least average workloads plus a bit.
Clive Longbottom, Contributor,
The enterprise application has been the mainstay of computing for a long time -- organisations firstly bought highly specific applications that were then packaged together as customer relationship management (CRM), enterprise resource planning (ERP) and supply chain management (SCM) "solutions". This has led to a marked redundancy of function throughout the data centre -- siloes of not only data but of code that is there to find details on customers, suppliers, inventory or whatever.
The knock-on effect from this is that information may differ from system to system, leading to errors in how data is interrogated and making an organisation's decision making capabilities suspect. It also has the obvious impact on technology resources -- the redundancy of function means that the effective utilisation of the resources is even lower than is measured through standard means.
Attempting to remove such redundancy through approaches like enterprise application integration (EAI) has proven complex and constraining, as well as expensive. Even then, it has proven difficult to isolate and turn off the redundant functionality, so efficiency gains in resource utilisation have been low. Further, the software vendors aren't interested in whether you are using a function or not -- you still have to pay for it, so there is no option for saving at that level.
Finding a solution
The emergence of Web services and service-oriented architectures (SOAs) has provided a basis for this to all change -- and virtualization, along with cloud computing, provides the last pieces of a jigsaw that should change the way that technology as a corporate resource is viewed.
If a standard business process is taken as an example, there will be a customer who needs to find an item, order it, pay for it and get it delivered. In today's world, this crosses the Web e-commerce site, the CRM, ERP and SCM systems, and may well involve a few more applications as well. For the process to run smoothly, each system has to have its own data stores, and they all have to be linked through some means so that each one can share its knowledge as required.
Turn the problem around: each customer, supplier and product is unique. If these were created as discrete records, anything that needed details on them could call the records associated with that item. It now becomes a case of is the customer already known? If yes, get all payment, delivery and other details from the master record for that customer. Is the product in stock? Look at the master record for the product -- if it is in stock, great; if not, then get the supplier master record and see if and when they can deliver one. Each stage of the process becomes a requesting or a responding service. If we change the process, the information flows are changed -- there is no "application" that has to change.
Further, take a more immediate example. A retail operation wants to open a new retail outlet. It can interrogate the details of its existing customers, looking for all those who have bought off the Web and where they live. It can call a Google or Bing maps function -- served externally -- and map the data directly on this and identify the hot spots where the largest density of customers are. No need for a mapping application, no need to run it inside the corporate data centre. Want to see if it would be better to open up a concession inside someone else's store instead? Use the same data, compare with existing shopping habits from the target store's systems and see where the hotspots are now. This is not using an application; it's making the most of the data that has been growing across organisational value chains.
Attempting to remove redundancies through approaches like enterprise application integration has proven complex and constraining, as well as expensive.
Clive Longbottom, Contributor,
The virtual data centre makes the most out of this through being dynamic and through being able to apply the resource to the workloads that need it. From an application point of view, any resource calculation is aimed at ensuring that the application can meet at least average workloads plus a bit, while in many cases they are engineered to meet expected peak loads.
What's next for applications?
A functional approach means that resources can be allocated to deal with data as the main focus. Functions tend to be small pieces of code that require small amounts of resource. If a lot of a function is required, it is far easier to allocate more resources on the fly. The function can be delineated, the basic workload of the function is known, its performance limitations (such as whether it needs a lot of CPU, I/O or storage) should be more easily understood and, as such, be able to be dealt with in a flexible manner.
In essence, the days of the monolithic application are numbered. Sure, we're in a "long tail" situation: the applications will be around long after I'm dead and buried. However, the clever organisation will start to plug holes in existing application functionality through calling discrete functions provided either internally or externally. As more functions become available, the "composite" application takes over: the application that is built on the fly to meet the needs of the individual, group or organisation as it battles to deal with ever-changing market forces.
The virtual data centre is a necessity to enable the functional composite application: the composite application is a necessity to be ultra-competitive in tomorrow's markets. As IT, it is your responsibility to advise your organisation accordingly. Prepare for function!
Clive Longbottom is a service director at UK analyst Quocirca Ltd. and a contributor to SearchVirtualDataCentre.co.UK.