Steve Singer <ssinger at ca.afilias.info> writes:
> Our dba team tends to run the slons from central admin servers and I 
> think they are happy with how it works out.

The case that I know has turned out badly is when slon processes were
running in a network segment distant from the DBs they manage.  

We'd occasionally have network glitches between sites (consider "virtual
LAN tunnelling" and such), and that would make those particular slons'
connections stop; it would typically take a couple hours (based on
TCP/IP timeouts) for the database to conclude the connection was dead,
and close it automatically.

Until that happened, replication would cease, unless someone went in and
terminated the useless connection on the DB server by hand.

If the slons managing nodes at "site A" run at "site A," and the slons
managing nodes at "site B" run at "site B," that sort of problem
normally doesn't take place.
-- 
output = ("cbbrowne" "@" "ca.afilias.info")
Christopher Browne
"Bother,"  said Pooh,  "Eeyore, ready  two photon  torpedoes  and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"