I thought that I’d drop a challenge on TK’s doorstep to see if I could distract him further from his authoring responsibilities, and a question leapt unbidden into my mind: How can parallel query ensure consistency when it can’t detect dirty blocks in the buffer? God ‘un, eh? See parallel query reads the data files directly, so how does it know whether there are any dirty blocks in the buffer or not?
Not only did this strike me as a very deep and weighty point directly relevant to some issues I’ve been grappling with recently, but I couldn’t think of a way of working out what the mechanism might be either.
Well these technical issues are often like good cryptic crossword questions: you can’t work out the answer for the life of you but once you hear the answer it all sounds very obvious.
Read the thread starting somewhere around here.
I’ll be back, Mr Kyte … oh yeah, I’ll be back …