To utilise a “Rich Client” – ie a bespoke application on the endpoint may resolve the issue of connectivity problems for users by allowing them to work without connection to the service for long periods of time, however it removes several of the fundamental drivers of cloud computing.
Firstly it reduces drastically the scalability of the solution across multiple thousands of endpoints. A significant driver of cloud computing (and its building block virtualisation technologies and SOA based architectures) is to reduce the reliance on endpoint applications and move towards an easily scalable solution which provides rapid expansion and deployment capability. Changes to code and application functionality are done once in the cloud based model without the need to push a client change over a period of time to multiple endpoints, something which takes a significant amount of time for the large enterprise. It also removes the reliance on the endpoint operating system and its patching state (which is more likely to be intermittant and inconsistent across the enterprise).
Secondly it dramatically increases the administration overhead on any enterprise looking at the BAU support processes around such a solution. With a client on every endpoint, which by their very nature are mobile and not connected in office locations with support mechanisms locally available, changes and issues become difficult and time consuming to resolve.
With the ubiquitous nature of mobile comms (be it HSDPA, EDGE or GPRS) and the fact that well design thin client cloud based apps actually transmit little in terms of data over the network a true centrally located thin client cloud solution is still better suited to the large enterprise.
In terms of a multi-thousand enterprise endpoint estate the volume of users affected by inconsistent communications covereage is a very small percentage of the overall enterprise user estate. As such the overall TCO models for cloud (if done properly) far outweigh those of a thick client per endpoint with its associated administration costs.
Anon.

Thanks for your well thought out comment. I agree, in our experience the “Rich Client” approach does allow users to continue working through protracted periods without connection, and in the real world these exist more often than you might think. I can't agree that the Rich Client approach “reduces drastically the scalability of the solution” as it enables scalability through Over the Air updates, rapidly deploying changes as required through profile management policies. Cloud computing is accessed though a thin client running on a device with an operating system. It's just as dependent as a remote device on its operating system and its operating system patching state.

Your second point hits the nail on the head! You've followed my blog so you know our interest is mobile workers, mobile devices and mobile solutions, its our passion. What you call “the ubiquitous nature of mobile communications” isn't ubiquitous! Our experience with thousands of mobile devices is that HSDPA, EDGE, or GPRS are not available everywhere, ergo in practical terms more intelligence is required in the hand of the user to control unexpected events, such as lack of network. Few organisations would take the draconian view that the small percentage of users affected can be ignored, they can't! This is where the TCO of any Cloud computing solution, if calculated with the cost of data loss and it's recovery, when applied to a mobile solution is usually higher than the Rich Client alternative.

Terran

Hi Terran,

I have enjoyed reading your informative and insightful article above and would like to add further to it if I may.

The points raised above are very pertinent and valid where applied to mobile applications that are dependent upon 3G/GPRS/EDGE connectivity to pass access and exchange data. However, I do feel that there is a great deal of value in cloud based enterprise applications such as Warehouse Management Systems, In-Store Stock Management solutions, etc. that can utilise WiFi connectivity for web access as sufficient resilience can be designed into the wireless network.

The benefits that cloud applications bring in such applications are very valuable in providing a high level of system availability, version control, data security and much more.

I look forward to hearing your thoughts on this.

Thanks for your interest and your comment, Mick. Sorry for taking so long to respond, it's been hectic lately.

As you know, our focus is on the mobile solution, so you'll understand if my comments are a little biased. But you're right there certainly is value in 'Cloud computing' based enterprise applications, I just can't help wondering if the concept of a central processor communicating with remote access terminals isn't what people of our generation used to call “Mainframe” computing. That said, there are many applications which require multiple input devices with central reporting, including; financial accounts, CRM and ERP.

As you say, system availability and security are among the benefits, with single version updates and controlled maintenance schedules offering further benefits. Although I see WiFi as historically less secure than a wired connection, modern encryption has overcome many of the pitfalls and it remains the most flexible carrier available.

Terran