- Image via Wikipedia
As the adjacent picture shows, even as highways increase in size, they become congested and speed and predictability decreases. Highway departments have added HOV lanes and emergency vehicles are supposed to have priority, but to a large extent, your experience on the highway is dictated by the traffic around you. Travel early in the day or late at night and it may be more predictable.
This appears to be an age old question; certainly one that is coming up again as 802.11n is introduced. This same question was asked when Ethernet was growing from 10Mb to 100Mb, to 1Gb and 10Gb. Even as 10X bandwidth increases were made available, the need for QoS was evident – even more than previously thought.
Early networks were interesting and provided basic communications, but their throughput and predictability were not considered robust enough to run business operations. As performance increased, IT teams envisioned what new productive applications could be implemented with the performance. Manufactures also understood the importance of networking so the number of client connections grew sharply. The problem began when all clients attached to the network would utilize the available bandwidth, usually at the same time of day, 8 to 5. Just as with highways, the network would become congested during the prime time, making the network unpredictable for critical applications.
Enterprise IT departments wanted to allocate bandwidth for specific application, rate limiting certain application to ensure the business critical application needs would be met and be predicable. As enterprises deployed business critical applications, there were certain applications which took precedence over general email, file transfers, and web traffic. Critical applications could be MRP systems, SAP, Oracle, or voice over IP. The need for QoS had enterprises turn to ATM networks for their backbones to achieve their goals.
Ethernet evolved from a shared medium to switched and introduced QoS into the standards to manage such application environments. Enterprises have endorsed these QoS mechanisms in their networks today. However, the QoS transition from wired to wireless clients is not as robust, so the question remains; Will 802.11n meet the QoS needs for mission critical applications?
Although 802.11n, at 300Mbps, enjoys a 6x performance advantage over 802.11g/a at 54Mbps, RF is still a shared medium. The proliferation of Wi-Fi enables devices will range from dual-mode PDAs, to Netbooks, and laptops, and will include a mix of 11g and 11n clients for a long period of time. This mix of clients, as well as the various applications needs to be managed within the enterprise to produce a predictable environment which ensures business critical application requirements will be met.
For enterprises to endorse WLANs as their primary access technology, WLANs need to create a QoS conduit to bridge the wired Ethernet QoS over the RF, all the way to the client, ensuring predictable delivery with switch-like low latency and delay, and allowing each client to view the network as if it had its own road.
