by John Wobus  Upgrading Your Network Backbone Is it time to upgrade? The first question you must ask is "Is a backbone upgrade required?" Before you begin this project-or instruct someone else to do it-address the motivations for the upgrade. You may find that the backbone upgrade can or should be delayed. Reasons to upgrade fall under the headings of performance and capacity issues, reliability issues, cost issues, technology issues and even publicity issues. (The latter may seem a strange, or even bad idea, but many network upgrades seem to be fueled in part by the publicity they will provide. Sometimes that publicity gives tangible benefits. Such issues can be a challenge to the technologist.) The alternative to a major upgrade is to buy a little more of what you have already: Expand a bit more rather than replace. We'll look at the reasons to upgrade: To address poor performance Altho ugh this is seems like the most obvious reason to upgrade a network, it requires you to establish that current performance is poor. This is a difficult task as people tend to adapt to the performance provided by the systems they work on. You must decide when performance is poor enough that an upgrade will pay for itself in increased productivity. To gain superior performance This year's superior performance might be next year's norm. You might consider your network's performance normal, even if other companies would consider it poor. It is quite possible that increased performance will yield benefits that pay for the upgrade. Hasn't all progress in computing and communication been about finding ways to harness the latest leading-edge performance to increase productivity. To provide the performance required by new applications Your organization may be planning to deploy applications -such as large file transfers, networkwide backups, systemwide application launches, increased network storage, multimedia applications including video and/or audio, or document imaging- that will make greater demands on the network. In general, applications that transfer text to desktop machines for display can be deployed widely without affecting the network load. On the other hand, just a relatively small increase in graphical applications or applications that transfer files could have a huge effect. To provide the performance required by planned expansion Your organization's plans might include providing network service to additional people or sites. To accommodate physical changes to your campus/buildings Your clientele is moving to another location or a building is being renovated. As the purposes of physical space changes, the networking needs of the location change. To increase the reliability of the network Outages can be costly in terms of productivity. But network equipment vendors are constantly producing more reliable products. Ensu ring networking reliability might be reason enough for an upgrade. To switch to a more cost-effective technology or vendor Over time, the cost of buying and maintaining of newer equipment with the same function decreases, presumably reaching the point where the resources dedicated to maintaining the old equipment are more than enough to capitalize and maintain newer equipment. To eliminate old and outdated equipment Age in itself is one reason to upgrade. As equipment gets obsolete the vendor support it disappear, forcing your own support staff to take up the slack. To switch to a technology with more of a future In the computing and network business, future technological directions are laid out years in advance, and it is often possible to adopt a new technology in line with the stated plans of the industry. The danger is in migrating to a technology that pan out. Unfortunately, only hindsight can reveal which changes advanced quickly and which lagged or stalled. Weighing against the reasons for upgrading are the resource costs-in terms of actual money and staff time, the decreased productivity associated with any network down time and the risks associated with changes and new equipment. Predicting the future Many network upgrades are motivated and guided by predictions of the future needs of the network. But how do you make these predictions and how can you ensure that they are accurate? If you know the network will be serving twice as many people doing exactly the same thing your current users are doing, you are a lucky manager. More often you'll have to gather information and make educated guesses. For instance, when deploying a new application, you'll hear various predictions of the bandwidth it will require, often by people inexperienced in measuring network load. These seers may have a poor idea of the scale and not know your current network load well enough to guide you. The application vendor may provide some cla rification, but its testing may not have been in a real-world environment. Others using the same application might have the information you need, if they are savvy enough about what is happening on their own networks. Perhaps you are simply accommodating ongoing growth in the size and use of the network. If you've been keeping measurements of network usage, you can graph long-term trend lines. Or you might just hedge your bets by over building. Waiting for the technology to settle down One trap that sites fall into is to put off upgrades in anticipation of the technology "settling down" to the point where no further major changes are apparently imminent. Unfortunately, such a time never comes: There is always something new, often predicated by gaps in the current technology or by attractive applications of the future. Although it is might be in your best interests to hold off adopting some brand-new technology, waiting for the hypothetical "right year" is not an option. Next Upgrading Your Network Backbone What Do You Need To Know? What Sort Of Upgrade? Evaluating Proposals Additional Issues Updated March 14, 1997 Print This Page E-mail this URL