I agree with Vlad's comments.

Computers (including servers) get cheaper by the day, but programming (labor) costs are holding steady or increasing. And every time you increasing the complexity of an application, it raises the cost of development, maintenance, etc.

Assist the decision makers in doing a Cost/Benefit analysis of the proposed project. There's no question the costs will be higher moving to n-tier. The level of developer skill is also raised, making it harder to find developers in the future to maintain the application, which factors into long-term costs.

I'm not trying to talk you out of the project, because it can be done successfully, but I have no doubt the initial cost of development and deployment will far exceed the cost of a few servers, especially since it sounds like those servers are already in production.




>Hi!
>
>>The rational behind this proposed move is purely that the client, which is a cash and resource starved UK national health authority, doesn't want to support servers at different geographical locations. Their concept is that an SQL base system at HQ linked to other locations without servers will be less of a drain on their resources. My system is the only system in use throughout the authority which uses local servers.....but then again it is the only system within the authority that meets the local users's needs!
>>
>
>Nice point. Unfortunately top management rarely look to this fact... ;(
>
>>My view is that any move away from local servers to a centralised SQL type system will result in a much slower data transmission time, especially as I understand the data links in place are very slow indeed. It would also result in hugely increased support time, given the proven unreliability of the data links as well.
>>
>
>This could cause only increase in development time and complexity of the application. Current technologies and concepts allow to build applications that way that work with acceptable speed and reliability despite unreliable connection. However, implementing them accurately require a LOT of effort, learning, designing and thinking. Support effort for databases decrease, though, - just because you will have only one server instead of multiple, so no data synchrinization is needed. Little of benefits, a LOT of work...
>
>>I have been asked, however, to look into this option and the feedback from yourself and from the others who responded has been incredibly useful. I shall be using it in support of my argument against going to a centralised data source, which will be a relief to me and my (small) organisation as it would put major demands on us which we could do without right now!
>>
>
>You can say them - "too much work for a little benefit". Indeed, great effort require great money, that could be the main argument against implementing this change ;)