On Wed, Feb 17, 2010 at 4:20 PM, Dave Cridland <[email protected]> wrote:
> Another distinction between the two approaches is what the core aims are -
> in PSA-style, it's to provide resilience between servers, whereas in
> KD-style, it's largely to reduce redundant message traffic from being
> repeated redundantly repeated.
Well, yes, mostly - in PSA-style you get resilience from arbitrary
points on the 'net, in KD style you still get resilience for your
users (and downwards, if you're looking at some tiered structure, as
you might set up if you've got increasingly remote setups) - and
through the SRV records, if you choose to deploy them, also resilience
from arbitrary points.
You could imagine a setup where you've got a flow something like
(apologies for ascii art)
earth
/ \
spaceship spaceship
/ \ / \
shuttle shuttle shuttle shuttle
...
clients
Where each server is shadowing for the entity above, and in that way
if contact between your spaceship and earth goes down, you still get
connections to your spaceship, and your shuttles, and all the clients
using those. From the clients pov, they just believe they're
connecting to [email protected].
/K
