It is pretty interesting to see the current evolution of SharePoint from an agile technology curve perspective. Briefly speaking if we consider a classical software project where we deliver services to a customer in a certain time frame, most project have got three typical phases:
1. Set up: most technology requires some kind of a setup phase, like installing the software, setting up the basics for the environment and usually buying licenses. This phase can be characterized with increased efforts (setting up effort and license fees), on the contrary relative few direct user requirement are covered. This phase is typically realized by infrastructure teams.
2. Efficient agile phase: in this phase user requirements can be delivered fast and efficient. Use cases that are delivered here can really be regarded in an agile way, like without explicit specification, consulting with the customer on a daily basis, having sprints and delivering use-cases on a weekly basis. This phase is typically realized by small project teams of consultants and developers, usually working on site at the customer.
3. Architecture limit: Once the architecture limit has been reached further use-cases can only be delivered for a huge cost, usually needing to extend the architecture itself. In this phase agile cooperation can not really be efficiently supported, instead classical requirement engineering and product development are required.
Figure 1. General Agile characteristic.
Platform versus product.
Comparing a classical product with a rapid application development framework from the agile technology curve there are major differences. Firstly setting up a product means delivering a lot of end-user functionalities as opposed with a framework where setting up usually means only a minimum scenario usually without covering too much end-user requirements.
Secondly, a product offers limited possibility for efficient agile development. There is certainly some possibilities to set some parameters but the situation can be very easily reached where the product itself has to be further developed, A rapid application development framework is however planned to realize a long effective agile phase in which user requirements can be fast and efficient realized.
From the architecture limit point of view there are not too many differences. If the limits reached than either the product or the framework itself has to be further developed that practically means slow and hard core software development.
Figure 2. Product versus framework compared.
Briefly formulate a framework is usually a general application development framework on which you can realize a lot different user requirements in an agile way with the help of a some IT consultants. A product is practically something ready, if you like it you get from the shelves and pay at the cash desk, if you do not like it, you look for another product. Extending the product itself usually realized only by the product manufacturer.
The last couple of years of SharePoint development was more or less going very strongly to the direction of Office 365. Even if there is still at least one version on premise, this version supposed to be similar with the cloud one. Besides, it seems that on a long run more or less the Office 365 will be the standard perhaps only on the cloud.
From an infrastructural point of view Office 365 is similar to a normal cloud solution, it is practically a product. You have a couple of well defined services that you can order and buy right away and after some configuration the set up phase has been delivered. As it seems to be a good thing for the first side, however it is more or less the one size fits for all principle.You can build up perfect data models based on lists and document libraries supposing that you did not reach data view limit however as soon as you have reached there is gonna be difficult to extend, you are practically at the architecture limit Certainly provider hosted app and storing everything in an Azure database is a possible way, but then you should consider implementing the provider hosted app, the logging mechanism, the user interface, the integration of standard Office365 lists, search and workflows... it is a lot of work for practically zero functionalities. On the contrary with a classical SharePoint on premise environment there are some possibilities , like extending the data limit on the central administration, integrating BCS with SharePoint Designer and SQL Database, having some Visual Studio Development and developing BCS with like Oracle...
Similar limits are to banning the public pages, constraints on the cross site information, problems on the brand new workflow and so on.
Based on the agile characteristics it seems that SharePoint is appearing more or more in the direction of a product and not as a rapid application framework. There are a certain number of predefined components that you can by and configure but if your requirement is coincidentally not supported than you can buy another product or practically finance platform extension. Actually the naming conversion shows this direction as well: it is already Office 365 that implies that SharePoint is a similar product as Word, Excel or PowerPoint and not a Rapid Application Framework. Similarly future names of App are actually AddIn-s that also points into the direction of rather being a Product as a framework.