The Fast and the Furious – Wireless Networks Smashing it Up Like it was 1999

They say that history repeats itself but who would have thought that the progressive world of Information Technology networks would become victim to this adage. In the mid 1980's the Ethernet 10Base5 standard connected networks using coaxial cables. Network designers of the day were only too aware of the limitations of the technology. It was a shared medium! Only one computer could transmit or receive on the wire at any one time. So as the number of computers increased so did the possibility of packet collisions. And they occurred by the billions on busy networks wreaking havoc on network performance. As a general guide, shared networks would grind to a halt from approximately 40% utilization due to the excessive number of collisions caused by that amount of traffic.

Collision domain, which is the term given to a segment of the network that is prone to collisions due to multiple senders transmitting simultaneously, was a term heavily utilized by network engineers until the 1990s when Ethernet switches appeared on the scene. Ethernet switches eliminated the single collision domain problem by isolating collision domains to individual ports giving other computers connected to the switch the freedom to transmit whenever they liked.

The shared medium design problem was well and truly cracked – until now. Enter the wireless network. A trademark of technology and economic efficiency the wireless network still shows tremendous promise as an integral part of the future of network transmission mediums. Why spend money on cabling when we can send packets across the airways?

But just like a dormant bacteria, wireless networks brought back a nasty design conundrum. The shared medium is back, and this time it's wireless.

Fortunately, many companies who tackled the shared Ethernet problems of the 80s are still around today. Some progress has been made with later versions of wireless access points to relieve the problem by separating backhaul and client access frequencies. This change certainly helps but essentially the problem still exists. Particularly on the client access segment.

So what can we do about it? The best solution to date is to go back to the good old shared media design principles we thought we shelved in the 1990s. Be aware of the number of clients accessing the wireless segment, add multiple segments if necessary to reduce the population on each segment. Note the kind of traffic and applications that will run on the segment and educate users appropriately. Be aware of other wireless characteristics that may aggravate performance problems such as interference and attenuation. Use appropriate antennas both on the access points and on the laptops.

Just like the good old 90s, decisions by modern engineers can now impact wireless network performance dramatically. Which is a bit of a relief, switches had made life too easy and we needed a challenge, right?