Ryan Malayter writes:
> On 9/14/07, Tim Lundström <[EMAIL PROTECTED]> wrote:

>> There have been discussions about a distributed monitoring system in
the
>> past. What happened? Are there still plans for a distributed
monitoring
>> system. With todays system an entire country's pool servers can be
>> kicked out of the pool due to one failing link between the national
>> exchange point and the monitoring server (OK maybe far fetched but
all
>> servers using the same ISP).

> I would volunteer to host a monitoring node in Chicago, Illinois,
USA.
> But obviously there has to be some coding work done to support that.
> Could it be be as simple as mathematically avergaing the score on
each
> distributed monitoring server?

One purpose of distributing the monitoring servers is so that poor
network connectivity/asymmetric latency at one doesn't mess up
the scores everywhere. Diversity of network paths and endpoints
is the goal, not letting a bad single network path drag things down!

To achieve this goal with just a small number (less than 5?)
monitoring
servers, I think you would take the best cumulative
or maybe individual score, so that one monitoring site being screwed
up with regard to connectivity doesn't drag anything down.

After you get more than 5 monitoring servers, then having one or two
with temporarily poor connectivity isn't such a big deal and maybe
you do something like throw out the worst and average over the
rest.

Tim.

_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to