Andy Mulholland, CTO, Capgemini
Budget season will be starting soon and of course all options for saving money will be required and onto the table go all suggestions. Virtualization is probably under way, so what about the other options that new technology is offering? The popular three are moving to a shared service center, data center consolidation and Business Process Outsourcing, BPO. And circling around these are the issues of software-as-a-service, security, and some very practical issues of exactly what is involved in each. Why is cloud included? Mainly because it is over-used, often wrongly, to describe anything and everything, and as such it's not really a helpful term. In fact it's more of a hype point adding confusion, and this blog is a 'back to basic options' piece.
Exploring options seems a good idea except it can be time-consuming and expensive if you don't know what you are looking for, and it's also pretty concerning when considering changes to enterprise applications. Consolidating enterprise applications looks a good bet, and certainly with ERP it usually is a good plan to consolidate the number of instances in use around a global enterprise. But what about other applications in use around the enterprise in various departments which do roughly the same thing? Assuming you can get the political agreement of everyone, and that's a big assumption, then it's usually less easy than you might suppose unless the selected new single shared application has a good set of migration tools. The 'gotcha' is in the databases, often different products, but almost always the non-obvious point is that the data models are too different to be readily reconcilable.
The time, cost and risk of the resulting migration is usually enough to discourage the project and to prove the projected savings might take longer than is acceptable after paying for the migration. The practical answer is to start off with considering migration tools and options at the beginning, and use this to figure out the choices, rather than the classic evaluation of the applications to establish a best, acceptable choice, and then look at the difficulties in deploying. Checklists on application migration and on data migration are available online to help anyone planning these moves.
It's usually easier to take the route of data center consolidation to handle this issue as at least 95% of applications can be operated on virtualized servers, so moving to fewer data centers, and consolidating in the remaining data centers is a simpler solution. The US Government has widely published details and the results of its huge consolidation exercise . The real long-term issue in this is not the traffic and use patterns of today, it's the patterns that are building up, and the long-term issues about high or low cost energy and the telecommunications zone. Specialist advice on the long-term outlook for energy and telecommunications coupled with emerging markets and the shift towards running as many 'services' as applications can all change the obvious choices of today. Data center planning in terms of location and use has never been so critical as at this conjunction of so many changes. CIO magazine had an interesting article on the five pitfalls of data center consolidation a while back, but frankly the horror story quoted is just bad planning.
BPO has become really popular over the last couple of years, but what and where to consider applying? Funnily enough it's back to some of the same issues as mentioned above; invoicing has long been considered as ideal for BPO and is a perfect example of what makes a good process to outsource. There is no competitive differentiation in the ability to prepare an invoice other than cost, and the data on an invoice is, in most countries, defined by accounting rules so is structured in a manner that a third party specialist can immediately use. Volume and expertise in operations allows the specialist BPO provider to achieve both better costs and to remove the need to support a vital service by employing in-house expertise in case it is needed.
BPO works in other areas where the ownership of the process doesn't provide any differentiation but is a business necessity, and can be said to be defined and structured by clear circumstances, such as HR, or Purchasing. So the real question in all three examples is less about the visible part of the actual application, and more about the underlying data model and how far this can be 'accommodated in any change'. Of course the flip side of this is planning ahead, and the explosion of data in various forms in the so called 'big data' generation ahead is think very, very carefully about databases and data models!!
So there are some good moves to get the cost down, even simplify operations and improve service levels, but the low hanging fruit has mostly gone and it does need to be approached in a skilled manner to gain from the experiences of those who have gone before! And most of all it needs to have some real consideration of future requirements that could look very different as more and more enterprises move to embrace technology for new roles around their markets and work practices with new devices.