Managing The Network |-------------------------------------| | Paper presented at CAUSE92 | | December 1-4, 1992, Dallas, Texas | |-------------------------------------| MANAGING THE NETWORK Jerry Martin The Ohio State University Columbus, Ohio ABSTRACT Once you have strung the wires and created a network, or multiple networks, you soon find that there are many other tasks that need to be performed. Compounding this problem is the various perceptions that different groups of individuals have of the network. From the user's point of view they want an easy to use, reliable, inexpensive way to utilize the network. The people who wire the network are concerned about keeping track of where all the wire goes and fixing breaks. The people who design the network are concerned about gateways, protocols allowed over the network, and traffic throughput. All of these individuals may not be interacting with each other for the good of the total network performance. These problems grow bigger and bigger as the network grows and departments want to connect their computers to gain access to the campus network and all of its capabilities. This paper will concentrate on managing the look of the network to the end user. Another paper, "Building an Effective Organization to Run a Campus Backbone Network", presented by Claudia Jordan at this conference, will concentrate on the organizational aspects of managing a network. ---------------------------------------------------------------------- What should a network do? In my opinion the most important thing for a network to do is to be invisible to the end user. This may seem a bit one sided and crass, since a massive effort has gone into the design, installation, and monitoring of networks. However, if you stand back and look at the purpose of the network, it is to facilitate the flow of information. It needs to leverage the knowledge of the end users of the information. It is to that end that we must always keep in mind as we design and implement our networks. Just as the telephone company does not emphasize its digital transmission and switching capability, we should not be trying to discuss TCP/IP when talking about our networks. We should be trying to "manage" the network as an information providing tool. The underlying wires and protocols are the foundations needed to provide the means of this information providing tool, but they are not the end product. We have traditionally concentrated our efforts toward the technology of the network, viewing it as a physical entity, rather than how it is utilized by researchers and secretaries. We have created this thing called the network, we now need to nurture its growth in the open environment in which it has flourished. How do you "manage" something as diverse as a network? Perhaps the word manage is a poor choice. As Grace Hooper so aptly said "You manage things, you lead people". As professionals in this industry we need to keep abreast of the technology and provide the leadership to steer us into the future. That is a wimpy politician type statement that does not get down to just what it is that we need to do. So, as Ross Perot has said so many times recently, lets get right down to the issue. Establish an organization that is set up to act on the directions needed to implement your institution's goals. Use existing programs to make the network as easy to use as possible. To make a network system work effectively, you need several basic things. + It has to work reliably. + It has to connect you to the resources you need or want. + It has to be usable by mere mortals. + It has to have consulting support. + It has to be fast, and able to grow as usage increases. + It has to be reasonable in cost and supported from the top. In order to have a working network, it needs to be viewed as a system of interlocking parts; that, when viewed as a whole, has a nice "look" and "feel". The network system has to have all of the technological components to make it work, but it also needs the end user to be able to use it effectively. Central computing centers are no longer the center of computing, as they were 10 years ago. Distributed computing is no longer something we talk about as a possibility, it is a fact. How we react to the next phase, that of providing leadership in the integration of the network to the needed data, has a strong relationship to our ability to exist in to the year 2000. Providing a network support system. A network has many separate components that have traditionally been separated organizationally. You may well argue that a person out in the trenches, training end users, does not need to know where and how the campus backbone routers are organized. If you get a call from and user setting up a LAN on the same side of a router, you soon find out differently. The network organization has to be set up so the knowledge of the network design experts is disseminated to those who need to know. We need to use the very network we designed to get this information to remote individuals who can not attend all of the informational meetings providing this information. We are in a distributed environment, with limited time and budgets. We need to utilize every technique we have available to us to distribute information to end users and network administrators out in the departments hooking up LAN's. One component of the network organizational structure is the function of a Network Information Center. This may not be a separate department, but it must be a part of the network function. This is the area I am most familiar with, so I would like to highlight some if its functions in the next section. The Network Information Center (NIC) On a national network level, the Network Information Center is normally involved in providing information on how to connect the campus networks to larger regional or national networks. It is usually not involved in providing end user information. On a campus, or local level, the need for end user support is very much needed. Because this function is so important, a NIC was set up at Ohio State University shortly after the campus network was installed. We dedicated a full time person to assist in documentation and developing training materials. In a two year period over 63 documents were created to meet end user needs. On a local basis, you may have several mainframes, and a multitude of microcomputers, running various communication links connecting everything. Providing all the documentation for each of these links is not a trivial task. The tact we have taken at OSU is to break documents into small sections, each addressing one specific task. For beginners, there is a "Getting Started with the Network" that describes what network is, and how in can be used. At the next level, we have the "How to connect" type document describing connecting to local and remote computers as well as file transfer techniques for each computer configuration. These documents are available in printed format and available 24 hours a day. The most popular are broken up by major topic and provided on-line via a Gopher interface. Even with all of the documentation in the world, you still have to provide training. In most cases, we provide training on networking in the same fashion that the documents are organized. For beginner, there is the "What can Networking do for you" type of workshop. There are e- mail workshops for those primarily interested in sending e-mail, using the various e-mail systems available. Finally, there is the "How can I effectively use the resources available on the network" type of workshop, for those who need to go beyond the basics. There is the consultation function for those specific questions that have to be answered "right now." For instance, are you prepared to answer these questions? + What is the difference between the Internet and BITNET? + How to I get the e-mail address of Joe Blow in England? + How do I get my Wordperfect document sent via e-mail? + How can I get access to the computer that allows me to display the pseudo stress vectors of the South Pacific for March of 1989? These type of questions require a different level of knowledge than many of our traditional consultants are trained to provide. We need to be anticipating these needs and developing the tools to provide this type of information easily. Using existing resources. There are many resources that you can use to arm yourself to meet these challenges. They are available on the network and mentioned in several recently published books. Gopher - A program running on many types of computers that allow very easy navigation of the network. This is a must have. If you are not running a server for your facility you should be thinking about it seriously. Archie - A program that can search for files available via anonymous on the network. WAIS - A program that can search specialized data bases provided on the network. Pop Mail - A program to allow reading mail using the native PC's interface, such as the Macintosh's point and click techniques. Some recent books of interest. Zen and the Art of the Internet - a Beginner's Guide by Brandan P. Kehoe, PTR Prentice Hall, ISBN 0-13-010778-6. BITNET for VMS Users by Michael A. Moore and Ronald M. Sawey, Digital Press, ISBN 1-55558-094-7. Crossing the Internet Threshold an Instructional handbook by Roy Tennant, John Ober, and Anne G. Lipow, Library Solutions Press, ISBN 1- 882208-01-3. CICNet Resource Guide, edited by J. Paul Holbrook and Christine S. Pruess, available from CICNet, Inc. 2901 Hubbard Dr., Ann Arbor, MI 48109-2016 for $27. The Whole Internet User's Guide and Catalog by Ed Krol, O'Reilly and Associates, Inc. Problem tracking. As the network grows so does your user base. Problems can enter your organization at many points. Even if you have a central point for reporting problems, they will not always reach that location. A problem tracking system can provide the needed information on what is going wrong in your network. Armed with this information, you can make better decisions on what action needs to be taken to fix it. It can also point out reoccurring problems. Finally, it can act as a central location for information. If your modem pool is down, anyone in your organization can call the central location to discover what the problem is, and when it can be resolved. Putting it all together. I would suggest we should not try to manage our network, but rather to lead those involved with the network on a trail to the future. To do that, you need to develop a computing plan for a short period of time, from 3 to five years. You should include: + A comparison of your institution with institutions of similar size on key measures, such as computers connected to networks, or computers outdated and needing replaced. + Specify goals of your network. + Specify what requirements you feel need to be meet to goals. + Specify the possible consequences of meeting or not meeting your goals. + Show how you would implement each goal, and the relative importance of each goal to the well being of your institution. This plan needs to be developed with top management, so they will at least be aware of the problems that you will be faced with in the short term future. Money will no doubt be a problem, but at least everyone will know the consequences of not meeting specific goals. All of these things need to be managed as a whole, to operate as an organized entity. Even with tight budgets and more work, failing to plan for the future may very well mean you may not have a job in the coming years. If you can not meet the need for change and provide an essential information service, you may well be a target for the budget axe. We can not afford to do the same type of things we did as a centralized computing facility. We need to keep up with the needs of our end users.