On 07/14/14 09:35, Dave Cramer wrote:
> Well distance pretty much translates into high rtt. I am talking about
> Toronto to Mumbai or the equivalent

There was a substantial change in communication in version 2.2 where the 
data is pulled via the COPY protocol now as opposed to using a cursor 
and small FETCH operations. This reduces the number of round trips 
required especially when there is a backlog and the system starts 
grouping events together.

You can relatively easy reproduce a WAN scenario in the laboratory using 
WANem, whe Wide Area Network Emulator.

http://wanem.sourceforge.net/

The thing comes as a Knoppix image that is rather easy to set up in a 
virtual machine and use as a router between vlans. You can impose 
bandwidth limits, packet delay (round trip), packet loss, packet 
corruption, packet duplication, packet reordering, idle disconnects, 
random disconnects ... you get the picture.

In my tests a cluster based on 2.2 is mostly affected by bandwidth and 
packet loss due to the delays that can introduce, leading to unused 
bandwidth. High round trip times (delay) are not much of an issue.


Regards,
Jan




>
> Dave Cramer
>
>
> On 14 July 2014 09:29, Andrew Sullivan <[email protected]
> <mailto:[email protected]>> wrote:
>
>     I don't think "distance" is the problem.  "High rtt links" might be
>     one, though: you can have cases where you time out or never catch up.
>
>     --
>     Andrew Sullivan
>     Please excuse my clumbsy thums.
>
>      > On Jul 14, 2014, at 9:16, Dave Cramer <[email protected]
>     <mailto:[email protected]>> wrote:
>      >
>      >
>      > How well does this work. Does anyone have some real world
>     experience with this ?
>      >
>      >
>      > Dave Cramer
>      > _______________________________________________
>      > Slony1-general mailing list
>      > [email protected]
>     <mailto:[email protected]>
>      > http://lists.slony.info/mailman/listinfo/slony1-general
>
>
>
>
> _______________________________________________
> Slony1-general mailing list
> [email protected]
> http://lists.slony.info/mailman/listinfo/slony1-general
>


-- 
Jan Wieck
Senior Software Engineer
http://slony.info
_______________________________________________
Slony1-general mailing list
[email protected]
http://lists.slony.info/mailman/listinfo/slony1-general

Reply via email to