This is the home of the gd2 network monitoring infrastructure. Its purpose is to give a real scale proof of concept of the design emerged from the Network Monitoring research group in the frame of the FP6 CoreGRID project.
Roughly speaking, we introduce an architecture for carrying out on demand network monitoring. Monitoring requests are expressed according a XML Schema, and are routed from the host that requests the monitoring data, to the one that carries out the monitoring activity. The monitoring activity is carried out by monitoring tools which are configured as plugins: we provide a plugin for a tool (nprobe, by Luca Deri) which is able, for instance, to monitor FTP transfers. Data extracted from the monitoring activity are routed to the requesting host using a UDP-supported stream.
More information is in the CoreGRID reports TR-0033 (2006), TR-0087 (2007), and TR-0126 (2008)
The software that implements the components of the gd2 infrastructure is available as DEB (Ubuntu) and RPM (Scientific Linux) packages. You can use the repositories listed below for a "plug and play" installation for i386 architectures.
To install under an Ubuntu distribution add to your /etc/apt/sources.list file
deb http://www.di.unipi.it/~augusto/gd2/dist/debs/ ubuntu/
Next update your package registry and download the package with its dependencies:
sudo apt-get update sudo apt-get install gd2-agent
To install under a Scientific Linux distribution create a file name gd2.repo in the /etc/yum.repos.d/gd2.repo and edit it with:
[gd2] name = gd2 baseurl = http://www.di.unipi.it/~augusto/gd2/dist/rpms/sl enabled = 1
Only in case you need to rebuild the packages (for instance, to support a different architecture) we provide a workbench to develop Ubuntu and Scientific Linux distributions. A simple howto is included in the package (sorry, it is in italian). The GD2 Workbench includes the sources as well as java compiled versions of everything needed to perform a real installation.
Your network should be partitioned into NM (Network Monitoring) Domains (first hit, follow DNS). For each each Domain you need to identify one (or more) NM Agent(s), and one (or more) NM Sensor(s). Sensors should be located so that they are on the path of the relevant traffic.
One LDAP server should be located, that will serve the whole network monitoring infrastructure.
Your infrastructure is ready to use.
In order to use it you need to have clients: we provided a minimalistic GUI for this, that you can install wherever you like (your laptop in the airport is OK) using the gd2-client. It is useful to understand the operation of the infrastructure, and you can capture messages to see their format. We envision workflow submission and monitoring applications with the gd2 client embedded, controlled by the application itself.
Now, if you have read up to this point, and have serious interest, it's time to go and read the real docs: start with TR-126 and proceed backward...
If you want to make reference to this product in your papers, please use the following citation:
@INCOLLECTION{Ciuffoletti:gd2:2008,
author = {Ciuffoletti, Augusto and Marchetti, Yari and Papadogiannakis, Antonis
and Polychronakis, Michalis},
title = {Prototype Implementation of a Demand Driven Network Monitoring Architecture},
booktitle = {LNCS from CoreGRID Integration Workshop (to appear)},
year = {2008},
address = {Hersonissos (Greece)},
month = {April},
url = {http://www.di.unipi.it/~augusto/papers/cur_2008_b.pdf}
}
Augusto
Ciuffoletti