This is more of a general question really, so I wanted to see if anyone had any opinions on the matter. I work in an environment where we had a lot of applications; and we're heavy users of SQL Server across these applications.
I was wondering if anyone could share their experiences on application testing across various SQL compatibility levels. The basic rule of thumb we have followed to date, is that we develop and test against the highest compatibility level of our current minimum supported SQL version.
For example if our minimum supported version is SQL 2012, then we will ensure the level is set to 110 for dev, test and into production. However, as a result of this, we are slower on moving / increasing the level which means we can miss out on initial features that come with using a newer version. We adjust this slightly depending on the application.
However, it's our older applications, which can be quite large in terms of code and features. Given testing across all SQL compatibility levels could be quite a large / cost task on our older apps, then I would be interested to hear how other people have handled this. I'm off the opinion, that to know for sure, then you must test it as much as possible. But given the increasing release cadence of SQL server, how sustainable is this? For context, we have mixed manual and automated regression. Be interesting to hear other opinions/experiences on the matter.