The Reports

This page will house the performance reports I will generate. Each of these reports is generated from the raw data  I am currently gathering from my residence. Read on to learn how I have generated this data.

These reports will be released on a monthly basis when my life allows me the time. I started monitoring the network on the 16th June 2009 and have continued to maintain the monitoring ever since. This page will be updated with each new report so check back here for the latest information.

The reports are based on the output from the Ping programme directed at the first “hop” interface managed by Internode.

The Reports

Performance Report for June 2009 – actbroadband.net_June09_Report – Incomplete Month

Performance Report for July 2009 – actbroadband.net_July09_Report

Performance Report for August 2009 – actbroadband.net_August09_Report

Performance Report for September 2009 – actbroadband.net_September09_Report

Performance Report for October 2009 – actbroadband.net_October09_Report

How is this Information Gathered?

In early June I decided to do some monitoring for myself. I have a few unused hardware platforms lying around from my days as a Unix administrator so I grabbed one and built it up. For the techo’s out there I installed a version of BSD Unix for an Operating System and downloaded the latest version of an open source network monitoring software package.

I configured a single network test based on Ping. The network monitoring software has been configured to send a small number of ICMP Ping packets at regular intervals to gain a view of the network performance across an entire day.

The ICMP Ping traffic passes through Crace exchange on its way to Internode’s Point of Presence (POP) at TransACT House in Braddon**. The LNS interfaces/devices are the first that respond to Ping. Ping works at the Network Layer (OSI model Layer 3). All the Crace Exchange infrastrcuture operates at the Physical Link and Data Link Layers (OSI Model Layers 1 and 2) and can not be monitored without authorised access (which I will never have)

I have NOT used Simple Network Management Protocol (SNMP) and I have NOT logged into any network devices or infrastructure I do not own. It is important to stress that I am performing these tests from my residence using my own hardware and know how.

I used a second network programme called Trace Route to determine the network path my traffic traverses to get to the Internet. From the Trace Route I was able to identify the devices within Crace exchange my traffic traverses. The first interface which responds to Trace Route traffic is lns2.cbr1.internode.on.net. This has recently change to lns1.cbr1.internode.on.net.

I import the raw data into Excel and use a number of formulas and the charting functionality to generate the reports. The reports are then converted to PDF format for upload to this site.

** Thanks to Andrew for the clarification on Internode POP.

Use of Ping

Ping is a programme that uses Internet Control Message Protocol (ICMP) to determine the time taken to reach the destination IP address and return. More information on Ping can be found here.

Ping can be run from the command line and is provided out of the box with nearly all Operating Systems including Windows, Mac OS-X and all flavours of Linux. The Ping programme produces an output that shows two important statistics – Return trip time and packet loss.

The Return Trip

The Ping programme sends an ICMP request packet to a destination, the destination device then responds to the ICMP packet with a reply packet. The Ping programme tracks the time between sending the request and receiving the reply, the time taken between the two is considered the return trip time.

The return trip is considered an indicator of the level of congestion on a network. There are a number of devices between a user’s router and the ISP network router. As the volume of traffic increases through these devices, delays can be experienced as traffic is queued to gain access to bandwidth as it comes available. The greater the congestion the longer the traffic sits in queues.

Packet Loss

If an ICMP reply is not received within a certain period of time the Ping programme considers the packet lost. Packet loss is another indicator of congestion issues. The greater the congestion the higher the chances packets will disappear off the network. Packet loss is a major concern for a number of reasons, once packets start being lost within the network the traffic is retransmitted by the source. This traffic will get queued again and possibly lost for a second time, compounding the issue and extending delays.

The reports show the output of the monitoring data gathered from using the Ping programme.

Page updated 28th September 2009 11:00am – additional information in “How is this Information Gathered?” and clarification on Internode POP location.