UNINETT internalweb: Mping


Mping

Detailed description of the Mping measurement service

Jump to section: What is Mping? :: How is it done? :: Groups :: Responsetimes :: Reports :: Overview :: Traceroute :: Sorting :: Graphs :: Histogram

What is Mping?

Mping stands for Multi-ping and it is a system for collecting statistics over packet delay and loss in a TCP/IP network using ICMP echo. Mping uses the InterNet Control Message Protocol (ICMP) "ECHO" facility,to measure round-trip-delays and packet loss across network paths.

Each night our net-monitor server Stasi pings a variety of Groups. Click here to learn more about how this is done.

Mping is based on the C-source code ping.c from Mike Muus, U.S. Army Ballistic Research Lab with various modifications:

Median and percentile support added by Einar Fløystad Dørum, for Uninett A/S

Multihost support added by Vegard Engen, for Uninett A/S

Various fixes incl. y2k-fix by Olav Kvittem, for Uninett A/S

Multi-lingual support, by Frank Aune, for Uninett A/S

Software Status:

Mping is released under the following license agreement:

Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions
 are met:
 1. Redistributions of source code must retain the above copyright
    notice, this list of conditions and the following disclaimer.
 2. Redistributions in binary form must reproduce the above copyright
    notice, this list of conditions and the following disclaimer in the 
    documentation and/or other materials provided with the distribution.
 3. All advertising materials mentioning features or use of this software
    must display the following acknowledgement:
    This product includes software developed by UNINETT.
 4. Neither the name of UNINETT nor the names of its contributors 
    may be used to endorse or promote products derived from this 
    software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY UNINETT ``AS IS'' AND ANY  EXPRESS OR 
 IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE  IMPLIED 
 WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR  PURPOSE 
 ARE DISCLAIMED.  IN NO EVENT SHALL UNINETT OR THEIR CONTRIBUTORS BE 
 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 
 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT  LIMITED TO, PROCUREMENT 
 OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR 
 BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 
 WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE 
 OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, 
 EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

If you are interested in using Mping on your network, please contact: Olav Kvittem. The installation is easy, and an extensive documentation is provided for your convenience.

Bugs :-)

Displays a png-image even if it's 0 bytes (Should instead generate new) - Fixed by Frank Aune, 20.11.2001

Features to come (hopefully)

Switch from one group to another on a specific date/period.

Zoom feature on graphs / histograms (from day <-> week <-> month)

Multi-lingual support will improve over time (switch during browsing is planned)

More statistics could always be gathered...

How is it done?

On ping consists of X (typically 100) ICMP echo-packets to a 'Gateway'. An ICMP-packet is a special type of IP-packet designed to be used for inspecting the network load. The most known tools which uses ICMP-packets, are traceroute and ping in Unix® and Unix-like OSes.

Mping then logs:

Lost packets.

Minimum Responsetime.

Average Responsetime.

Maximum Responsetime.

We have several techniques to make sure that Mping collects "valid" data:

Mping doesn't send more than 10 ICMP packets per second -> Meassured data is independent from the time of meassuring.

Mping doesn't send all ICMP-packets to one 'Gateway' at the same time, rather Mping tries to spread it out -> Avoids temporary network characteristics.

Mping starts the ping-job at asyncronous times. We use a Poisson-distribution -> Avoid periodic network variance.

For more information about this subject:

RFC2330 Framework for IP Performance Metrics

The Groups we currently ping

Mping ping several different groups of hosts. Each group has it's own host list. Please note that some hosts are not online 24h a day, so evidentally, they will sometime get 100% 'Packet loss' (eg. Wireless networks)

The different groups we currently ping, are:

Uninett

Bibsys

Trondlost - Trondheim Wireless

Trofast Servers

The World

Also, we provide statistics for the ATM network from 15. January 1999 to 20. January 2001. The ATM network is, however, outdated so we don't ping the ATM-group on Mping any more. See the Mping frontpage to view the statistics for each group.

Responsetime-statistics Uninett provide

This is an overview of the different reports and histograms we provide for various groups, for which we ping. The date or timeperiod is always presented in the left column. You can either view 'Responsetimes' or 'Histograms' for specific months or for specific days.

(Hint: If you are presented with three columns of URLs, you are in the monthly overview :-)

To view 'Responsetimes' and 'Histograms' for a specific days, click on the 'Daily Reports' for the corresponding month.

To view 'Responsetimes' for a specific month/day, click on 'Responsetime' for the corresponding month/day you want to view.

To view 'Histograms' for a specific month/day, click on the 'Responsetime distribution' for the corresponding month/day you want to view.

How to read 'Responsetime reports'

In this section you will learn how to read the 'Responsetime reports' generated by Mping. The following information is presented in a report:

Links: Pretty obvious...

Help? This document.

See also: Use this to navigate between the different statistics for the current group.

Sort by: Click on the desired sorting method for the Gateway list. To learn about the different sorting methods, see 'Sorting'.

The 'Overview' option.

Note: The selected sorting criteria appear as plain text in the 'Sort by:' menu, and not a clickable URL.

Running a 'Traceroute':

Gateway (route): By clicking the gateway name, a traceroute to the corresponding gateway is executed. See Traceroute for more info.

Understanding Responsetime and it's appearance:

The 'Responsetime' for a corresponding 'Gateway', sorted by 'Median', 'Max' and 'Variance'. For descriptons of the different sorting methods, see 'Sorting'.

By clicking the 'Median' value for a corresponding 'Gateway', a graph is presented showing 'Responsetime' and 'Packet Loss'.

Understanding 'Responsetime distribution' and it's appearance:

There are four subcolums: <25(ms), <50(ms), <100(ms) and <200(ms). They each show how the 'Responsetime distribution' (in %) is divided into the respective categories.

Understanding 'Packet Loss' and it's appearance:

'Packet Loss' for the corresponding 'Gateway', sorted by 'Avg' and 'Max'. For descriptons of the different sorting methods, see 'Sorting'.

By clicking the 'Avg' value for a corresponding 'Gateway', a graph is presented showing 'Responsetime' and 'Packet Loss'.

Overview

The 'Overview' option under 'Sort by:' is especially usefull. You have the same sorting criteria here as you have under 'Responsetime reports'. But rather than presenting alot of numbers and percentages, the 'Overview' section has a text based display of the 'Responsetimes'.

Responsetime: The grading here is based on the average 'Responsetime' (ms).

Average Responsetime (ms) Grading

< 50 Good

< 100 Avg

> 100 Bad

No response from 'Gateway' No data
Packet loss (%): The grading here is based on the average 'Packet Loss' (in %).

Average Packet Loss (%) Grading

< 0.1 Good

< 0.5 Avg

> 0.5 Bad

Average Responsetime (ms)	Grading
< 50	Good
< 100	Avg
> 100	Bad
No response from 'Gateway'	No data

Average Packet Loss (%)	Grading
< 0.1	Good
< 0.5	Avg
> 0.5	Bad

How is a Traceroute executed?

The Internet is a large and complex aggregation of network hardware, connected together by gateways. Tracking the route one's packets follow (or finding the miscreant gateway that's discarding your packets) can be difficult. Traceroute utilizes the IP protocol `time to live' field and attempts to elicit an ICMP TIME_EXCEEDED response from each gateway along the path to some host, eg: 'Traceroute' displays the route packets take to a given network host. (taken from the traceroute man-pages)

A line instead of a number in the Min Median Max column indicates that there are no available polling results for this period. Note: The responsetimes presented with this traceroute, is taken from Uninett's database.

By clicking a 'Min (ms)' number in the list for a corresponding 'Gateway', the following graphs will be drawn:

Round-Trips for 'Gateway'

Packet-Loss for 'Gateway'

For more information, please see 'Graphs'

Description of 'Sorting' methods

This is a brief description of the different sorting methods Mping use:

The 'Gateway' list is sorted by:

Machine name

Median: The median value is the middle value in a set of values. Half of all values are smaller than the median value and half are larger.

Standard deviation: How tightly all the various samples are clustered around the mean in a set of data.

Packet loss (default)

'Responsetime' is sorted by:

Median: The median value is the middle value in a set of values. Half of all values are smaller than the median value and half are larger.

Max: Maximum 'Responsetime' for an hour or day.

Var (Variance): The variance is a measure of how spread out a distribution is. It is computed as the average squared deviation of each number from its mean.

'Packet Loss' is sorted by:

Avg: The average amount of sent, but not responded, ping packets.

Max: Maximum 'Packet Loss' for an hour or day.

Graphs for 'Packet Loss' and 'Round-Trips'

Two graphs are drawn:

Round-Trips for 'Gateway'

Packet-Loss for 'Gateway'

You can easily navigate between days by using the 'See also:' URLs. Note: A straight line in the graph doesn't nescessarily mean a linear increase or decrease. It is more likely that there are no existing polling results for this specific period.

The data can also be presented as a 'Scatterplot' by selecting this view. Note: If the graph exceeds the maximum value on the Y-axis, the plot is rounded down to the maximum available value.

Histograms showing 'Responsetimes'

The histogram shows the responsetime distribution for a given number of hosts. The axis are divided into:

Number of Hosts

Average Round-Trip

The Histogram view can be very useful when trying to get an overview of the netload.

nav-drift@uninett.no

2001-22-06

Startpage