We have had the Asia-to-North-America case; it was fine, so long as we had
sufficient bandwidth.

Latency wasn't great, but Slony actually copes with that quite well.

I think we had more troubles with reliability of links (e.g. - they were
passing through enough intermediaries that occasionally connections would
get deranged) than anything else.

Pre-2.2, it was crucially important for the slon servicing each node to be
local, otherwise the round trip times between slon and database would be
troublesome.  The shift to using streaming (e.g. - copying sl_log_{1,2}
around via COPY)  gets rid of that particular round trip issue.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20140714/d57b9ebd/attachment.htm