Florian G. Pflug <fgp(at)phlo(dot)org> wrote:
> I believe the hard part of implementing true serializability is
> not the actual SSI or S2PL algorithm, but rather the necessary
> predicate locking strategy.
>
> So I think checking how InnoDB tackles that and how much of it's
> code is invovled might give a more realistic estimate of the
> effort required.
Apologies, Florian. I never got your email, and somehow got the
impression from Greg's reply that you had led the discussion into
overly technical areas for this thread. I dug up your email in the
archives and found that the above was the entirety of your comment,
so I'll reply to that now, rather than my misapprehension. :-/
I agree that most of the work will be in the predicate locking area.
You're probably right that if I could isolate the lines of code
required for that in InnoDB it might get me to a more accurate
estimate of lines of code. And as Greg points out, it is bound to
be a bit more complex in PostgreSQL than in InnoDB.
At this point, though, I'm just looking for an "order of magnitude"
sort of estimate of complexity, since even a totally accurate line
count wouldn't do more than hint at the really important metrics --
how much money it will take, and how much risk of destabilizing
existing code there is. A certain amount of work is required to get
an accurate handle on those. That work may start soon, but we're
not there yet.
Thanks for the input.
-Kevin