Then there are things that PL/SQL can do – but not any better than Java, such as generating HTML. When you are thinking about scalable solutions, you could and probably should wonder whether using PL/SQL and therefore database resources, for producing HTML – especially the none-data driven HTML, provides the optimial, most scalable solution. Kyte is a big fan of HTML DB and MOD_PLSQL in general and I have used a lot of MOD_PLSQL in developing the Repository Object Browser (pka ODWA), but I really do not believe it a very scalable architecture. Perhaps if one database acts as data server and another as PL/SQL application server you have a more scalable layout. Personally, I’d go for the most efficient way of gathering the data to use for a webpage (for example using a PL/SQL based API that collects all data, in Collections or ref-cursors) and returning it to a Java based middle tier that sticks it into fancy HTML.