[smokeping-users] Any downsides to running less probes more frequently?
alter3d at alter3d.ca
Fri Jul 25 18:44:53 CEST 2008
David Tomic wrote:
> What is the detail that I'm losing in the packet loss/latency data by
> sending fewer pings?
> Is it basically just that [for example] a single dropped packet will
> count for 20% packet loss with only 5 probes being sent, vs only 5% with
> 20 being sent ... or is there something else involved that I'm still not
> quite understanding here?
It is partly the granularity (e.g. 5% vs 20% for a single lost packet).
This will tend to skew your loss statistics higher than they actually are.
However, the 'smoke' in your graphs also depends on sufficient sampling
in order to show a good representation. With 5 results, you simply
aren't collecting enough data to show how your network latencies are
distributed around (in terms of above/below) the mean.
Keep in mind that this tool is designed primarily to see trends and
problem areas at a high(ish) level -- it's not meant to drill down to
see e.g. why you get loss at 43 seconds past each minute.
Unless you absolutely need to get alerts within ~60 seconds if a link
drops, it's generally better to have more samples with longer scan cycles.
More information about the smokeping-users