The good news is that the cost of bandwidth has come way down, at least during the 20 years I've been buying it. Most enterprise LANs give users fire hoses from which they take periodic sips. Five years ago, we asked ourselves if it was worth twice the cost to give users 100 Mbps of switched bandwidth rather than 10. Cisco convinced us the answer was yes -- only wimps buy 10-Mbps switches.
Today's LANs have bandwidth to burn, and most of it goes unused thanks to the powerful lessons we've learned about application development in the past 20 years. If you design applications requiring boatloads of network bandwidth (which is exactly what we did in the early days of client/server), you'll be delivering those applications only on your internal LAN. Given the geographically distributed nature of today's organizations and their value chains, that is simply unacceptable.
The Wireless Wake-Up Call
It may not be at the top, but facilitating mobile access to information is high on the priority list for most IT managers. Unfortunately, wireless bandwidth is still a precious commodity. Today's 802.11b WiFi LANs, with data rates of 11 Mbps and an effective throughput of about half that, are no fire hoses. But it's not that bad. We got lots of work done 15 or 20 years ago with 50 to 100 users hammering away at a 10-Mbps Ethernet LAN. And you can get a lot of work done on today's wireless LANs too.
But is 5 or 6 Mbps enough? In most cases, the answer is yes. Experts designing enterprise wireless LANs always start by asking two fundamental questions: How many people will access the network in a given area? And which applications will they be running? Answering the first question is usually just a matter of overlaying head count on a building blueprint. But answering the second question is considerably more difficult. Relatively few wireless LANs are installed with a clear understanding of how they will be used. Sure, industry research tells us e-mail/messaging and Web access are the most popular applications today, but what about tomorrow? Will we need more bandwidth in the future?
Of course we will. But IT industry economics force compromise between network application design and basic network services availability. It doesn't make much sense to invest millions of dollars developing state-of-the-art, bandwidth-hungry wireless multimedia applications if they won't run on the existing installed networks. And, indeed, plenty of bandwidth-constrained WiFi networks are out there, and more are installed every day. So the best bet is that truly innovative applications, those that can be sold, will generate value in existing environments.
Progress Without Compromise
Will we be forever constrained by the limitations of today's WiFi wireless LAN technology? Not a chance. First, improved QoS (quality of service) capabilities will let us more effectively allocate the bandwidth we have, enabling functional but bandwidth-efficient applications like VoIP. And new multimode chipsets, arriving later this year, will preserve legacy WiFi support while adding support for newer high-speed standards like 802.11a and 802.11g. Alas, it won't be quite as smooth as the migration from 10-Mbps to 100-Mbps Ethernet, but it'll be close. The economics and technical design of wireless systems will also improve, to the point where we may one day have the technical equivalent of "switched" access, and we won't be contending for network resources with the guy down the hall.
Meantime, you're making a big mistake waiting for the next generation of wireless LANs solely because today's products don't offer enough bandwidth. In networking, the next generation will always be better than what we have now. To let yourself be paralyzed by that fact may stunt your organization's progress and that's simply unacceptable.
Send your comments on this column to Dave Molta at firstname.lastname@example.org.