Best Practice to build smaller .dlls from large db

What would be the best practice for building separate applications by object domain from a single database that could be combined into a single deployed solution? For example have one project for Products and another for Customers/Orders, etc? In a proof of concept I ended up with a 50 Mb .dll which would be less than ideal for load performance. It took many hours to complete generation. Also, I have tried to find some guidance on system requirements to improve Radzen performance, specifically RAM.



Not sure I understand your question. The size of the dll doesn’t matter in a server side technology like server Blazor. Even if we ignore this fact the application dll from a Northwind with pages (114) for all tables and views will be less than 1MB in Debug and even less in Release (when deployed):

Thank you for the quick response.

My database is much larger than that with a couple hundred lookup tables. It took many hours to generate the pages, and like I mentioned the .dll ended up at 50mb, so that's my starting point here. Looking to see how I could build and test smaller bits of functionality without having to wait so long to test each change. So any new functionality introduced would have to build the entire solution each time, right, otherwise if I list already generated pages as ignored, that functionality will not get compiled. Just looking for best practice for building a very large solution.


How many pages you have? Thousands? I don’t think that this is practical for web application.

That's not uncommon for the applications I have worked on. That's why I asked about breaking things down. Having a UI to add or edit values in a lookup table is nice for an application administrator, but not critical to the solution and most users. It could be done with update scripts, or I could create a separate app just for system administration that would be quite complex on its own and leverage the same authentication/authorization. Just trying to get a feel for best approaches before I start making changes to my data model.


Then I suggest you uncheck all tables from the tree in the infer data source screen in Radzen. Then check only the tables you need for the application.

You can also control what pages to generate per table.