The Fog over Grimpen Mire: Cloud Computing revisited

| No Comments | No TrackBacks
| More
Last night, at a BCS meeting on the legal issues around Cloud Computiing, Miranda Mowbray of HP Labs opened a most informative discussion by presenting the material in a paper (see also the slides she used) which built on the analogy used by Bill Thompson in "Storm Warning for Cloud Computing" . Many of those present were either looking at Cloud services for their organisations or grappling with implementation.   

Miranda began with some excellent definitions

At their heart, however, was the difference between a shrink wrapped license adjudicated under the laws of the State of Washington (with liability for failing to deliver anything at all, let alone a promise of 99% or 99.5% availability, limited to $500) and an Outsouce contract that typically includes a hundred page Service Level Agreement with penalty clauses adjudicated under UK law for failing to deliver 99.999% availability, as well as a raft of other obligations under Data Protection, RIPA, Consumer Protection etc. etc. etc.

The discussion over drinks afterwards enabled me to test out the responses I have received over the past week regarding my questions on the bandwidth needed to make effective use of Cloud computing (whatever that is).

My quick summary is now:

The bandwidth needed to provide access to the Cloud is akin to that for thin client only if the applications are also akin to a well-designed transaction processing service running off centrally held data files (e.g. that for the on-line renewal of your driving license - or for ordering books from Amazon).

The bandwidth needed increases dramatically if you expect rapid turn-round when submitting or requesting documents or images (e.g. a local authority planning system or checking identity biometrics at a border control point). It also changes if you are planning to run office systems (word processing, spreadsheets etc)   

The BCS meeting was attended by many from the public sector, both central and local government, who were looking at the issues they need to address if they are to take Cloud Computing (including the aspirations for G-Cloud) seriously.

There are indeed massive savings to be made - but these come mainly from the reduced support costs and license fees that result  from software standardisation (c.f. the US DoD agreeing a single, centrally supported, secure subset of WIndows XP for all desktops).

Rationalising in-house data centres can also yield very significant savings, provided these are already accessed via fibre-based leased lines. So too can the agreed relaxation of current outsourced service delivery standards - to the levels offered by the Cloud suppliers.

These savings can, however, be rapidly wiped out if the improvement in speed and quality of response leads to increased demand for services which "earn" less than the fee to the Cloud provider. We should not forget that the "business case" for the on-line driving license renewal systems is based on the willingness of the middle classes to pay £2.50 for the privilege of not having to queue in a Post Office. If the authority or department has a fixed budget then improved service can mean that it runs out earlier in the year.

So, to return to my original question, what are the bandwidth needs for Cloud Computing?

- organisations that wish to strip out corporate overheads need fibre-based leased lines (capable of being wound up to gigabits and more) between their locations and the Cloud service provider. This is because they are on a one-way growth path, which will accelerate as response times improve and cost-per-transaction comes down. More-over they will almost certainly wish to benefit from migrating office systems, not just transaction processing, onto the Cloud.

- customers and end-users who use only text (including numbers) based transaction processing services, with limited amounts of data entry or security checking, do not need significant bandwidth.

This changes dramatically, however, if they are expected to transmit or receive documents or images.

Programmes to:

  • improve accountability by allowing voters to viewing government spend, planning applications, MPs or Councillors expenses etc etc.
  • strip out reporting and service delivery overheads by allowing residents to submit photographs of problems (e.g. Love Lewisham or Tidy Oldham) and for those repsonsible for action to post photos of what they have done
  • use Web 2.0 and social networking as part of social inclusion and service delivery  

let alone 

  • run the high definition imagery exchange required for applications like telemedicine,

depend on the local provision of the level of broadband access (fixed and mobile) akin to that already being rolled out around the Pacific Rim or across Scandinavia - let alone that planned.

I will not comment on the legal Grimpen Mire described by Miranda. Follow the link above to her paper and consider the issues that need to be addressed for the Cloud suppliers and their customers are to benefit more than their lawyers. 

 

No TrackBacks

TrackBack URL: http://www.computerweekly.com/cgi-bin/mt-tb.cgi/37893

Leave a comment

About this Entry

This page contains a single entry by Philip Virgo published on February 19, 2010 10:05 AM.

Drowning in Data Leaks v. Death by Data Protection was the previous entry in this blog.

More bottlenecks than a brewery: the UK Broadband Network is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Archives

Recent Comments

 

-- Advertisement --