From a pure sense of logic, cloud computing is the right choice for almost every situation requiring enterprise IT. Hardware can be deployed as and when required, used on-demand and the management of this IT is taken care of by the cloud provider. Taking this to the next level, IT leaders are increasingly putting Platform as a Service and Software as a Service at the heart of their enterprise IT strategy
But people seem to be attached to the idea of ownership, where up-front acquisition costs can be spread over a number of years. Is it cheaper to deploy in the cloud and pay an on-going monthly fee forever or pay upfront for, say, five years of IT hardware?
Computer Weekly recently spoke to Adrian Bradley, a partner in the technology practice at KPMG, about some of the pitfalls associated with going wholly into the cloud. In Bradley’s experience, the main reason why organisations struggle is often because they fail to keep costs under control. Profits can quickly be eaten away when organisations fail to track closely the costs associated with launching new software-powered initiatives that use public cloud services. In some cases, organisations find that the costs are so high that it is cheaper to bring IT back on-premise.
Simplicity not Complexity
Some organisations see a hybrid approach as a means for them to have the best of both the on-premise and public cloud worlds. Based on the requirements of the IT workload, with well-managed IT operations, it should be relatively easy to determine where best to deploy storage and compute. However, as IT expert, Junade Ali, points out in a recent Computer Weekly article, building an IT architecture for hybrid cloud deployment is far more complex than one that is purely cloud-native. There are far more opportunities for something to go wrong. Moreover, organisations developing hybrid IT still need to maintain inhouse IT expertise to maintain on-premise datacentre equipment.
Hybrid is probably not a long term solution. It should be treated as a stepping stone towards cloud-native IT. However, this does not necessarily mean everything is processed in the cloud. James Donkin, CTO of Ocado Technology, for instance, sees the use of heuristics on edge computing devices as a means to preprocess data streams. Such heuristics is able to “understand” what normal looks like. Abnormal data is used for cloud-based machine learning, to update the heuristics. For Donkin, an on-premise datacentre is a “squeezed middle” between the edge and the public cloud.