Hi Anthony,
If you haven't taken it, definitely think about Hyland's API class (or push your employer to think about it).
They'll spend half a day talking about connection management, which is a good starting point even if a lot of it doesn't make sense in some technical situations and some licensing situations. That'll be a running theme- many of the "best practices" you learn there shouldn't be applied mechanically, they should be over-ridden by judgement, experience, and your situation, but the class will be a good starting point.
The best possible answer to your question will depend on what the application your developing is, the context it runs in, maybe your licensing. But as a starting point-
If your app is making a series of related calls over a short period of time, you definitely want to do it under one connection. It will keep your code simpler, it will be much more efficient at runtime (creating and closing a connection takes significant time), and it may be important for licensing depending on your situation.
So- if you're doing something like running a console app that needs to find a set of documents, retrieve metadata, and make some changes, you would almost certainly want to maintain a single connection.
If you're building a web app and thinking about one user opening a page and performing a series of operations, the answer is "it depends"- you may be trading off runtime efficiency, licensing considerations, and code complexity. You're probably going to do something that equates to "one connection per user session".
If you're building a web app and thinking about many users connecting over time, you'll probably be creating many different connections- but it still may depend a little on how you're doing licensing and other factors.
