Casting a high-altitude eye over the SQL / NOSQL debate, two things immediately come to mind:
- The debate seems to almost always boil down to SQL vs Not-SQL (i.e. Whether or not SQL is broken and an alternative needs to be found)
- People are comfortable with SQL - it might not be ideal, and it handles a lot of things sub-optimally, but it generally gets the job done, even if you have to fight it into submission occasionally.
Apart from this panel discussion on GigaOM, I haven't seen the 'Not Only SQL' message really addressed. The question, at its core, isn't necessarily about whether or not SQL *as a whole* is on the way out (that's a separate debate), but rather whether non-relational alternatives (paid-for and free) should be embraced when they are better suited to certain applications.
I see a lot of people respond to the "Is SQL Broken?" question with responses like "It works pretty well in most cases...", "It does most of what I want it to...", and "I'm happy with good old SQL". SQL has evolved to work in a wide variety of instances, and it's carrying a lot of baggage as a result. I think it's fair to say that SQL is so widely used (and this explains why we're hesitant to give it up) precisely because it does most things reasonably well. But the downside of that flexible, work-around-able, kludge-able ubiquity is that there isn’t really anything that it’s designed to do brilliantly. So perhaps we do need to be more ready to employ some of these other data storage models when they are the right tool for the job at hand. Besides that, I'd say if it isn't broken, don't fix it. GoodOldSQL.
