When to purchase ‘cloud computing’ services?

A recent article from the economist indicates that it is more than just a hype and that even though the term may become passé, the cloud itself is here to stay and to grow. Cloud computing accordingly to Wikipedia is ‘a style of computing in which dynamically scalable and often virtualized resources are provided as a service over the Internet. Users need not have knowledge of, expertise in, or control over the technology infrastructure "in the cloud" that supports them’.

This makes cloud computing to me not so different compared to other hypes like Software as a Service (SaaS), ASP or any other service delivery models where application functionalities are made accessible over the network/internet from a third parties location. I thus wonder whether the buzz word ‘cloud computing’ will be replaced by another flashy term in the near term if it does not show the return predicted by the marketeers….(I for example hardly hear anybody talk anymore about ‘utility based computing’. Has that term been replaced by HaaS (Hardware as a Service) the brother of SaaS.

Google is a company that provides these kinds of services. The most well known is Gmail, but they also provide spreadsheet, wordprocessor and other functionalities. After seeing that Gmail went down for the second time in a short period (May 8th 2009 and February 24th 2009) I started to wonder whether ‘Cloud Computing is mature enough to be a service that should be used for ‘mission critical’ activities. But with companies like The NY Times, Amazon, Nasdaq, British Telecom, CSS and Activision using cloud services (most of them not using Google btw) I guess I am (partially) wrong.


Still, I believe that ‘cloud computing’ is currently only useful for a limited range of IT services as Clouding requires:

  • A constant internet connection. No connection no service, and as there have been several cases of some ship or other object destroying some internet cable (examples here and here), is this Clouding not an solution for business critical processes, but only for those processes than can be done manually for a while if needed.
  • The servers can be anywhere.
    You might not have enough control of your servers which might get you into a) trouble with the legislator for not complying to certain privacy regulations and b) performance issues and lower response times compared to connections with near/onsite datacentres. Especially certain manufacturing processes require very low response times in order to operate properly (e.g. chip foundries).
  • Functionalities are limited. You get typically an application which runs within a webbrowser and that limits features you can run (I think another factor is that too many features would make it too slow to work with as too much data would be travelling back and forth between the client and the server.
  • Less control over your data. Already mentioned a bit before. Cloud computing is a ‘black box’ and you have very limited insight in what happens behind your webbrowser. You can of course agree with the vendor to have all kinds of control in place regarding IT continuity, physical and logical security etc, but requiring many changes to the base model is difficult as the whole concept is based on: one-fits-all. Additional customisation also reduces the chance of achieving economies of scale, which affects the whole business case behind it.

To me Clouding is thus useful, but only in certain situations. Some of the prerequisites which have to be met are:

  • Business activities relying on cloud computing services can also be executed manually for a longer period of time (e.g. one or two days).
  • Business activities should not contain critical or very sensitive data (e.g. personal, intellectual property)
  • Cloud computing services should be based on relatively simple application functionalities like word processor, spreadsheets, email and some transactional business activities (e.g. keying in purchase orders)

Regarding specific sourcing/strategic risks of Clouding, I think the most important one is a more expensive lock-in. You cannot resource the services to another vendor or insource them are they are proprietary to the vendor. So if you want to get rid of this type of contract it is gonna be costly.

Comments

Popular posts from this blog

Beyond Two-Speed IT – Part 3

Beyond Two-Speed IT – Part 2

The importance to innovate as a supplier