Project

General

Profile

Actions

Bug #7959

closed

Dpinger - Probe Interval

Added by Lampros Papanastasiou over 6 years ago. Updated over 6 years ago.

Status:
Not a Bug
Priority:
Very High
Assignee:
Category:
dpinger
Target version:
-
Start date:
10/17/2017
Due date:
% Done:

0%

Estimated time:
Plus Target Version:
Affected Version:
Affected Plus Version:
Affected Architecture:

Description

Dear Team
We are working with WAN satellite connections limited in bandwidth data per month.
The maximum probe interval is 86400 msec that is very low due to the limited bandwidth.
We would like to be able to change this value to a maximum of 30 minutes in order to save bandwidth.
Thank you
Best Regards

Actions #1

Updated by Denny Page over 6 years ago

Hmm... this doesn't quite make sense to me.

A probe of 28 bytes every 86 seconds seems a minuscule cost. With a low speed uplink (64Kb), even a probe interval of 5 seconds is significantly less than 1/1,000 of the available bandwidth. And if you use a probe interval of 40 seconds, which is less than half the maximum, it drops to under 1/10,000 of available bandwidth. Or if you want to look at it in in dollar terms, at a rate of $.01/KB a 40 second probe interval is $0.59 per day. This doesn't seem a significant cost on a satellite link.

If you simply can't afford the 28 bytes every N seconds, it would probably be better to simply disable gateway monitoring completely. With a Probe Interval of 30 minutes, the minimum detection time for link down will likely be beyond practical use anyway.

If I've misunderstood something about your use case, please explain more.

Actions #2

Updated by Lampros Papanastasiou over 6 years ago

Many thanks for your input.
If the size of the packet is only 28 bytes ( 56 bytes calculating the reply as well ) yes it's not a big deal.
I thought that the size of icmp package is 74 bytes ( 148 bytes with the reply ) so the traffic per day is 144 Kbytes which - 4.5 Mbytes per month .
In some cases 5 Mbytes per month are expensive ( 50 - 60 US$) that's why i am trying to find a way to minimize the cost for the gateway monitoring.
It's not critical to failover in 84 seconds but is requited to failover in 20 minutes that is acceptable as time interval.
Thank you in advance.

Actions #3

Updated by Denny Page over 6 years ago

This would probably be better as a forum discussion...

A probe every 60 seconds is 40,320 bytes per day one-way, 1,209,600 bytes or 1.15MB per month. A probe interval of 80 seconds drops this to 30,240 per day, 907,200 or 0.87 MB per month one-way. Using the value of $10 per MB (expensive), and assuming the bandwidth cost is symmetrical (unusual?), the total bidirectional cost would be $1.73 per month.

To put this in context, a single load of https://www.google.com takes 15KB with a text tool. A browser (Chrome), will generally use over 4x of that (62KB). In other words, a full day of probes is still less than a single load of Google’s (empty) search page.

Unless the link is completely unused otherwise, there are probably better places to look for bandwidth savings.

Actions #4

Updated by Renato Botelho over 6 years ago

  • Status changed from New to Not a Bug

Please move discussion to a forum thread and if a real bug is found open a ticket with specific data

Actions

Also available in: Atom PDF