Network Topology and Performance Visualization for
the European Research Network
When there is a failure or performance degradation
in a trans-European network connection, it is rather
difficult for the end-users to obtain timely and detailed
information. Having access to this information is
of growing interest especially for the Grid projects
being established throughout Europe. Funded by the
European Community, the Joint Research Activity 1
(JRA1) of the GEANT2 project is aimed at measuring
the network performance of the GEANT network –
which connects the national research and education
networks throughout Europe – and providing meaningful
information to GEANT users. Experience gained from
a project at the Leibniz Supercomputing Centre is
going to play a vital role in JRA1.
Besides making available high performance computing
facilities, the Leibniz Supercomputing Centre also
runs the Munich Scientific Network linking the Munich
universities and other research institutions. This
network became the starting point for the development
of a network topology and performance visualisation
tool called Customer Network Management (CNM). After
the necessary data became available, the tool also
has been provided to the customers of the German Gigabit
Scientific Network (G-WiN) in April 2004. The G-WiN
is the national research and education network and
connects universities and research institutions throughout
Germany. The aim of the DFN (German Research Network),
which runs the G-WiN and funded the development of
the CNM tool, is to provide a similar tool for the
GEANT2 network. This work will be carried out as part
of JRA1 that has been launched in September 2004.
Figure 1: Customer Network Management tool for the
trans-European Research Network GEANT
A screenshot demonstrating how the tool could be
applied for GEANT is shown in the Figure. The network
topology is depicted as a tree structure on the left
as well as in the topology maps on the right. For
network nodes (i.e., routers, switches) and links
current information about their status and characteristics
are available. The status is indicated using colored
status codes. To quickly find errors in the network
hierarchy, a special network tree is provided showing
only elements having warning or failure status. In
its current version for the G-WiN the only characteristic
for the nodes is the number of forwarded IP packets,
while bandwidth, throughput, and utilization are provided
for the links. In addition to the display of current
values in the maps, it is also possible to get access
to past values for status and characteristics. For
doing so, statistics showing intraday, week, or month
charts are provided.
The network topology is discovered by using SNMP variables
defined for the use of the proprietary Cisco Discovery
Protocol. The status and characteristics are derived
from other SNMP variables. The client of the CNM is
written in Java using the JavaWebStart tool to launch
and update the applet. For the communication between
client and server, CORBA has been chosen, the server
itself is written in C++ and the data used for the
statistics is stored in a PostgreSQL database.
 top
|
index
prev