Veikko Brax

To my personal Web page

My history

My interest areas

Background of the IPANA project

The wide and ever-growing usage of Internet and the birth of new Internet applications ( e.g. telephony and video conference ) requires a smooth transition to network components providing both the old and new services. This has caused development of new label swapping  protocols for Internet routing, to make it possible to use  broadband technologies like the ATM in the transmission level.

Several solutions to do this task have been suggested by different parties e.g. Classical IP over ATM, ATM Forum LAN Emulation and ATM Forum Multiprotocol Over ATM (MPOA).

IPSILON Inc. introduced in the spring of 1996 the concept of IP switching, where the functionality of internet routers and ATM switches was combined. You maybe know, that e.g. Helsinki Telephone Company's Areena 2000 trial network is based on IP switching, and that the manufacturer of the IP switching routers - Ipsilon Networking, Inc. belongs now to Nokia Telecommunications.

After Ipsilon Inc. have several other parties proposed  their own versions including CISCO's Tag Switching, IBM's Aggregate Route-Based IP Switching (ARIS), Juha Heinänen's Switching IP Through ATM (SITA), Toshiba's Cell Switch Router (CSR) and the MPLS-workgroup's Multiprotocol Label Swapping (MPLS).  As the latest one is there now coming Augmented IP Protocol (AIRP). All of those technologies are usually called the IP swapping, and their basic technical ideas are also compared.

The basic idea behind the IP switching is to reduce the workload of the router by switching long-lived data flows, containing relatively large amounts of data, to separate ATM-connections. This concept is rather simple, but  contains also several challenging issues to be solved, e.g. Quality of Service, CAC and resource allocation.  Especially in my topic, implementation and optimization  of a large IP switched network opens a huge problem area.

Research done in theIPANA project

A lot of preliminary studies, theses and articles have been made in the IPANA project before I come to it, and I may refer to the respective pages and articles of my colleagues.

We also have in our lab a non-working research network, which is implemented by using  IPSILON technology. There are also multimedia traffic sources and the measurement equipment's available in the research network.

I have been working unto now with my Licentiate Thesis  "The scalability and design parameters of IP switched networks" but more about that later.

There are a lot of corresponding studies, and maybe the most interesting one could be  A Simulation Study Of IP Switching done by Steven Lin and Nick McKeown in Stanford University in April 1997.

My licentiate's thesis

My Licentiate's Thesis has a title "The scalability and design parameters of IP switched networks".  In this study I'm concentrating to a question, as how large could the private / public local telecommunication networks be  implemented by using the IP switching technology, when the services would cover the traditional Internet data services and also audio and video services like Internet phone, PABX and video conference.

When speaking about large networks, the importance of the topology is rising up. The number of different possible IP topologies are really increasing exponentially, when the IP network is expanding, and the optimization of the IP topologies is becoming more important. E.g. Ellen Zegura and Matthew Doarn have developed software tools to generate optimized random IP topologies.

The routing is also an important subject, when speaking about large IP networks. Several routing protocols have been developed, and each of them has some special area of use. If you like introduce yourself e.g. to the OSPF, there shoud be available a demo tool about the OSPF for that.

When simulating the operation of IP networks, the traffic definitions should be based on real traffic traces. There are really quite many traffic traces and studies based on them available in the Internet. It is possible to calculate profiles  for the most common protocols of the large IP networks.

The results of the study is the performance of the IP networks. That could be defined e.g. as the load of the nodes and the links. If the links are based on the IP swapping technology, the load of links could be understood as e.g. the sessions of Virtual Channel 0 and the number of Virtual Channels used for sessions. The load of nodes in turn could be e.g. number of sessions the node has to process in a time unit, or the number of signaling packets between nodes to manage the Virtual Channels, or the number of memory accesses the node has to do in a time unit.

The  results should always be validated too, and there are respective simulations done by e.g.  S.Lin and N.McKeown in Stanford University.

As the result I then hopefully have some information to answer to the basic question in this study.

In the case you would like to get more information about my study,  or you have any comments or questions to it, please send me  a message.  I would be very grateful.

Tämän sivun sisällöstä vastaavat, Veikko Brax 

Sivua on viimeksi päivitetty 02.10.1998 13:30
[TKK | Teletekniikka
? Hae => Anna palautetta!