mping

Availability and IPv6 support

Name	IPv6 support	Supported operating systems
mping	included	Unix, Linux, *BSD

Description

Mping is a system for collecting statistics over ping times and ping loss in a TCP/IP network. Mping is based on an earlier version used internally by Uninett, but for the latest releases most of the code has been rewritten.

Comments

The Mping service consists of two parts: The Mping client, written in C, and the web interface extension, written in PERL. Unless otherwise specified, in this document when referring to Mping we are referring to the Mping C-client Mping is a tool for measuring round-trip delay and packet loss, using the ICMP echo feature, in a TCP/IP based network. Multiple hosts - up to a compile-time define (500), both IPv4 and IPv6 at the same time - can be pinged in a round-robin order. At runtime, the user can set the wait time between each packet sent, number of packets sent and the size of the packets. For each host specified, information about packet loss and minimum/average/maximum response time is displayed. Mping can also display the collected data as median, cube-sum, standard deviation or 10/50/90-percentile at the users request. Several techniques are implemented into the Mping service, to make sure that the collected data is "statistically" correct:

Mping by default do not send more than 10 ICMP packets per second, thus measured data is

ndependent from the time of measuring.
- Mping do not send all ICMP-packets to one 'Gateway' at the same time, rather Mping tries to spread it

out in a Round-Robin fashion, thus avoids focusing on local temporary network errors.

Mping starts the pingsweeps at asynchronous intervals. We use a Poisson-distribution, thus avoiding

periodic network variance. Technique 1 and 2 are Mping C-client features, while 3 is implemented in the PERL web interface extension. A web interface is used for browsing the collected data and for generating reports, graphs and traceroutes for the different hosts we measure. The PERL code is modularized and easily extended to suite other needs. As an example, the language support is modularized and thus adding support for new languages is very easy. The mping backend performs tracroute to all destiations nightly. The router addresses are added to the ping host

ist and the route is stored. The we web-interface will show historic routes with historic data, thus monitoring outes over time. The traceroute display will indicate which hop that causes a particular high delay or loss.

tested and used in 6NET project

Usage information

Source of information

http://www.6net.org