Some interesting discussions always evolve around the topic of splitting strings. In two previous blog posts, "Split strings the right way – or the next best way" and "Splitting Strings : A Follow-Up," I hope I have demonstrated that chasing the "best-performing" T-SQL split function is fruitless. When splitting is actually necessary, CLR always wins, and the next best option can vary depending on the actual task at hand. But in those posts I hinted that splitting on the database side may not be necessary in the first place.
There were a lot of comments following my post last week about string splitting. I think the point of the article was not as obvious as it could have been: that spending a lot of time and effort trying to "perfect" an inherently slow splitting function based on T-SQL would not be beneficial. I have since collected the most recent version of Jeff Moden's string splitting function, and put it up against the others: