OK.

It really depends on the amount of data being processed. If you only have a few rows but each one contains a large amount of text it may take longer than a small amount text in many rows etc.

I would normally use the method shown in that link. That way you can parallelize the creation and you're not having to restructure existing blocks. Once the process is complete you'll have a clean table and you can drop the old one. If availability is an issue you should think about using an online table redefinition.

Cheers dude!