Home > Our Products > NetSkateKoban > Reports >

■ Installation Example of NetSkateKoban ? Tohoku University (Graduate School Economics Research Dept.)

Managing the network of a university is entirely different from that of an enterprise, right from the ground principles, says Dr. Ishigaki Masahiro of Tohoku University’s Graduate School Economics Research Dept. (center of the picture; on either side are network management staff)


■ Differences between University and Industry networks

-- On how many levels does a university's network environment differ from that of an industry?

This is absolutely based on popular opinion, but I think they differ most on the following points in application principles. For the sake of clarity I will explain in points.
  1. Several systems that are not possible to unify.

    The university network is an environment where various versions like Windows, MAC, UNIX, Linux, etc all coexist. Sometimes there are experimental machines connected for the purpose of research.
    In an ordinary business, since this would be detrimental to efficient work, OS and equipment would be largely uniform, but in universities PCs are used for research and studies, and therefore efficiency is not the priority.
  2. Outside PCs are permitted.

    Most business networks, for reasons of security, prohibit the connection and use of private PCs. But in a university, one can't afford to prohibit students and staff from bringing in their own PCs.
  3. Apart from students, connection by outsiders is also permitted to some extent

    For example, a researcher from another university brings a notebook PC for collaborative research ? the outside PC would of course be permitted to connect to the university network or say a PC used for a long time in an overseas study program returns to Japan ? it too would be permitted to connect to the network.
  4. Several usage policies exist together

    There are PCs used by one person all day in a static location, for example in a staff member's room; or like in the computer lab, where many casual users log in one after the other; or the office, where usage is quite similar to that of a business. Post graduate students' PCs go from the student's house, to the common research room. Many different usage policies exist in university.
  5. High turnover of network users.

    Simply put, in a university every year senior students graduate and new students join. From the network application perspective, it is as though one fourth of the network users are getting replaced every year. In businesses there will no doubt be people retiring and people joining, but most certainly not one fourth of all users.
  6. Network management is semi-autonomous if not autonomous

    Perhaps this is a characteristic only of Tohoku University's Economics Department network, rather than all university networks. In our department, we have a student volunteer organization called the Economics Department Student Mail Research group, which sets students' mail addresses and other network settings autonomously. It is something like having a grassroots management organization, apart from our general management branch. Even after listening to the methods of other universities, it appears this method is highly unusual.
■ Requirements of a network management tool for a university network environment

-- What are the requirements of a network management tool in an environment like this?

To put it roughly, 'registration of users who can connect to the network', 'identifying currently connected users and machines','detection of fraudulent connection','detection of and response to abnormal traffic (*)' We are presently carrying out these applications using NetSkateKoban.

-- What are the specific applications you are using?

I will explain each article along with appraisal of NetSkateKoban.
  1. Limiting users permitted to connect to the network

    In a university, users allowed to connect to the network would be students, graduate students, teaching staff, office staff ? basically students and university staff. Supposing a student wants to connect a private machine to the network, after submitting the application with the machine's information, like the MAC address etc, machine is registered in NetSkateKoban. In this way, an inventory of legitimate machines is created inside NetSkateKoban. If a machine which is not listed in the inventory is connected, it will be considered as a fraudulent connection.
    As mentioned earlier, every year one fourth of network users are replaced in a university. In a graduate school, there is an April batch and October batch ? with different course durations. There are also frequent visits by professors from other universities for the purpose of collaborative research. This will mean frequent large changes to the management inventory, and if the interface becomes difficult to use and are not intelligent, it will hinder network management.
    It is good that updating of NetSkateKoban is so easy. Even beginners can update it.
  2. Detecting users and machines connected to the network

    In our administrative office, we supervise the networks of the economics department and the graduate school. In a university network, even if theoretically there is a unified topology, it is normal for the physical places to be scattered. In the case of the economics department's network, it extends across B wing, part of G wing, fourth and tenth floors of I wing, fifth floor of F wing, three lecture wings ? the places are scattered. In an environment like this, it would be difficult to locate the physical area where the network abnormality broke out. However NetSkateKoban's ability to specify the connection area by switch and port comes in very useful in these cases.
    In the actual scenario, the Koban sensor is placed at the top of the tree from the source main switch, so if something goes wrong, it is possible to know immediately that something is strange with one of the machines connected on the fourth floor of I wing. To such extent it makes it very simple to determine the exact machine that is malfunctioning.
■ Ideal network management and future expectations from NetSkateKoban

-- We would like to ask, as a sort of survey, what would be the ideal network management for you, Dr. Ishigaki?

It's difficult to say what would be 'ideal', but assuming something went wrong in the network, it would be good to be able to know where it went wrong, how and what went wrong, and magnitude of damage. Since we are talking about an ideal system, it should be able to provide information without any prompting, in an easy-to-understand visual form when the event occurs. An ideal system wouldn't tackle the problem after it broke out; it would be able look for situations likely to cause trouble and nip them in the bud. If that can be accomplished to perfection, from the user's point of view there will be zero network problems. If that could happen the network would be able to function with as much stability and dependability as the electricity or water supply.

-- On a scale of one to ten, with your ideal network being a ten, how do you rate NetSkateKoban's current level?

I think it has come up to a pretty high level. If you use a pro baseball victory as a metaphor for the ideal case, NetSkateKoban gives the impression of 'soon to enter A class'. If nothing else, the vector this product is aiming for is pointing in the right direction.

-- What sort of hopes and expectations do you have for NetSkateKoban from now on?

Both security and network change drastically in this industry. I hope development of technology to deal with these changes is continued. Network is now an important part of the infrastructure of education. To protect the stability of this infrastructure, please continue to offer your superior products and services. I have high hopes.

-- Thank you for your time.


※ 'Detection and tackling of abnormal traffic' is a function of NetSkate Visualizer ※ Tohoku University (Graduate School of Economics Research)website
※ Date of interview ? September 2005

Copyright © 1997-2008 Cyber Solutions Inc., All rights reserved.