[MUD-Dev] On Lockless Threading and X/Open XA

[Greg Underwood]

Bruce writes:

> Greg Underwood wrote:
> > My initial thought would be to have the 2PC be the commit part of the C&C.
> > Any reason something that simple wouldn't work?
> > 
> > It's been a while since I did an in-depth read on either protocol, but I
> > don't think 2PC is necessarily anathma to a C&C approach.  IIRC, what it
> > buys you is the ability to rollback failed, partial commits.
> 
> I was somewhat wrong in my thinking of what phase 1 of two phase commit
> was.  It really is:
> 
>   Phase 1: Prepare.  Ask everything if it is ready and able to commit. 
> If
>     so, move along to phase 2.  If not, immediately rollback everything.
>   Phase 2: Commit.  Everything has said in the previous phase that it
> was
>     ready and able to commit.  At this point, everything can commit.
> 
> This works just fine with both C&C and some of the other schemes for
> deadlock-free multithreading, since you can query in phase 1 and at that
> point get an answer.  The only problem that might arise then is that the
> complete resolution might not be as immediate.  I will remember next
> time to not confuse lock acquistion with 2PC. :)

Wow.  I guess I really did learn something from those database classes. 
And I thought they were a complete waste of time.  Guess that'll learn me. 
;)

I actually was a little sad when I had to drop the final class part way
through (work... :P).  We were just getting into massive parallelization of
algorithms.  Pretty neat stuff, in spite of the professor's complete lack
of teaching ability.

-Greg



_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev