Thanks Henrik, Tomash,

In addition to the incorrect interpretations I think I'm mixing my
Ketama tests with looking at server failures.  The big win for ketama
(which is perhaps obvious) is scaling up -- adding a new server won't
trample the database as bad.


> If you want to do some useful tests that show you how the key distribution
> works, and the advantage of Ketama vs. the naive distribution, I would
> suggest these:

Thanks.  I may try, although I think you explained them well enough. ;)

BTW -- is the port also used in Ketama to map the servers into the address
space?  Or is it just the address?  Probably explains something that
didn't quite makes sense this morning.


-- 
Bill Moseley
[EMAIL PROTECTED]
Sent from my iMutt

Reply via email to