When to deploy the separation model?

[dazlunn]

I'm looking for some logical insight into "when" to deploy the separation model.

I am working on a new project which combines 3 existing FM systems (just the functionality not the actual systems) and we are writing the 'new' system from scratch.

This is going to be the largest data store I have ever worked with, in terms of population density of records (as opposed to number of tables). I typically work on smaller systems in general.

We are just at the pencil/paper stage at the moment but approximately there will be around 25-30 base tables (maximum) but at least 3 tables will hold around ½ million records (500,000).

The record population will only grow at around 10,000 records per annum so no problem there, I feel. What I really want to know is should I be thinking about designing this system using the separation model or will a single system suffice?

We also need to connect to a single external SQL table which sits right at the beginning of the data flow, i.e. an event in the external SQL table triggers/cascades actions through the rest of the system in one direction (not sure if I've explained that very clearly??). I don't know if this has any impact on determining when to use the sep' model but thought i should mention it..

I found lots of helpful info in the forum on "how to deploy" the sep' model but not "when" to deploy it, a sort of checklist if you like..

I appreciate any guidance you may offer..

p.s. around 100 fields maximum per table and that includes fields used for interface etc. Mostly single lines of text per field, such as postal addresses, names, short descriptions of parts, jobs etc --- the usual service industry suspects..