The Changing Face of Network Management - A Hungarian Experience
by István Tétényi
TEN-155 happened the other day. TEN-34 became history after 18
months. Research networking is still fashionable. People, organisations
have accustomed to the results of the Internet revolution. We
take ubiquitous, instant access granted. Is it not true? Just
the opposite. The situation is only seemingly under control. Networking
is like teenagers with long arms and legs and short clothes. Network
management is much like the caring parent that tries to meet the
demands. This article tries to summarise the results that we
have achieved to manage the largest slice of the Hungarian Internet
of several hundred institutions.
The Network Services Department of SZTAKI (NSD) is a small ISP
without the commercial chore of selling the services on the open
market. NSD has been managing Hungarnets IP network for many
years. (Hungarnet is the Hungarian Academic and Research Network
Organisation, http://www.hungarnet.hu) Some people define the
task that we do as network operation. Well, it is true on the
one hand but a bit less than the full image. It is true that we
operate a network infrastructure. On the other hand, this is ever
expanding in capacity and has got a wide range of technologies
from ISDN to SDH.
Working in a conservative and under funded environment sometimes
helps. You do not have to worry about the management approval
of requested funds and the bureaucratic procedures of procurement.
Although you will not end-up with glossy leaflets and boxes that
are a bit outdated in the day of delivery. On the other hand,
you might collect and develop a set of tools running on different
platforms and providing the very services needed. Traditional
network management meant that there were a few dedicated desktops
from where the operators and the real techies could check the
network status. The Internet revolution and the web explosion
affected network management. We are somewhere halfway through.
The big names (Microsoft, Cisco) are about to invade this territory.
Till it happens, there is room for home-made solutions. Below
you will find our approach and the results.
As in every modern organisation, the Intranet is the basic infrastructure
component that supports groupwork. We (NSD) do have a loosely
coupled set of tools put together in the recent years in order
to manage the networks we operate. Naturally, the tools alone
- without the people and procedures - cannot solve the tasks.
What we use at the moment is a Network Management Intranet that
was developed over the years. It has been refined and its functionality
extended many times.
Network Management Intranet = several functions (tools) that are
equally available for a group of people solving common tasks together.
A cornerstone of our approach is: there should not be any dedicated
console for dedicated network management. We have to have tools
that can be relocated in case of failures; we have to have a flexible
environment where at least a group of people can share the same
view of the network:
- Basic core tools: monitoring network reachability (pinger), monitoring
service availability (servers alive), network events administration
with ticketing (home-made ticketing software)
- Statistical tools: collecting interface traffic statistics (mrtg,
monster), collection individual IP statistics (home-made collection
tool)
- Extended monitoring tool: line database (home-made), interface
database (home-made), network database (home-made), web based
console for monitoring network reachability (home-made), service
level monitoring (home-made)
- Reporting - monthly reporting: service level reporting (home-made),
line overloads (home-made), traffic monitoring (home-made), ticket
statistics (home-made)
- Special tools: monitoring special events (collisions, temperature,
VC statistics), backup handling (home-made), paging (SMS) regularly
and in case of special network events.
Expected future developments
We plan to introduce netflow statistics collection this year to
have even more precise information about traffic characteristics.
Application monitoring will be extended. We try to strengthen
the database background for the tools we use. Security issues
have to be refined. We plan to migrate interface statistics collection
and visualisation from monster to rrdtool (mrtg, cricket). We
are looking forward to directory enhanced network management.
We are open to co- operate with anyone who truly believes in vendor
independent tools and network management.
Please contact:
István Tétényi - SZTAKI
Tel: +36 1 349 7532
E-mail: tetenyi@sztaki.hu