posted by Paul DeBeasi:
There have been many articles, emails, and blog entries written about the Burton Group report “802.11n: The End of Ethernet?”. As author of the report, I’d like to clarify a few things.
1. Burton Group takes pride in the fact that we do not do ANY vendor-sponsored research. We don’t even do vendor-sponsored white papers. Burton Group clients are predominantly very large enterprises and rely upon our research for unbiased, vendor-neutral research.
2. The report specifically looks at LAN access networks. Wired Ethernet will be used for many years for inter-switch trunks and data center connectivity. But, for LAN access, wireless is the future.
3. Speaking of the future, the report is looking at the state of the wireless/wired market in the 2 – 3 year time frame, not in 2007. In fact, enterprise 802.11n access points are not even shipping yet. I don’t expect RJ45 jacks to be ripped out of the walls. Instead, we are going to see wireless slowly spread and eventually dominate LAN access.
4. Security is a broad topic so let's break it down. 802.11 networks provide authentication, data privacy, and data integrity. If you use best practice wireless security (802.1X + WPA2 + AES encryption) you can deploy a wireless network with authentication/privacy/data-integrity that is just as good as a wired network. Sure, WEP was broken (and lots of companies still have WEP) but WEP is not a best practice. If you want strong wireless security, you have to use WPA2. It is true that wireless denial of service is a vulnerability but the real question is what level of risk can you live with and how much are you willing to pay for RF monitoring/mitigation? Also, don't forget that a wired network is not perfectly secure. Eavesdropping can occur on a WIRED network too (see http://www.governmentsecurity.org/articles/articles2/ACSI33-5-Emanations-and-Cable-Security.pdf_fl/).
5. If you use IETF RFC 2544 as the basis for benchmarking, the latency of an 802.11n AP and a Gigabit Ethernet switch is not very different (both are well below the ITU guideline of 150 milliseconds for VoIP). However, real world latency is impacted by traffic patterns and how the equipment is deployed. It is easy to create situations where wireless latency is high. For example, if you use 802.11b clients and configure the AP to allow connections even with very weak RSSI, then you are going to bring the network to its knees. However, many of the new 802.11n enterprise APs will have two radios. A preferred configuration is to operate one of the radios in "mixed mode" @ 2.4 GHz (ie, compatible with 802.11g/b) and run the other radio @ 5 GHz for 802.11n laptops only. In addition, the APs should be configured so that very weak RSSI signals are disconnected. Laptop clients tend to be "sticky" and hang onto their AP connection even when RSSI is weak.
6. It is certainly true that 802.11b/g/a APs can exhibit instability. By "stability" I mean wireless connection reliability. 802.11n and best practice network design can alleviate a lot of the problems. Much of the instability with 802.11b/g/a networks is caused by multipath and co-channel interference. 802.11n makes use of MIMO (multi-input / multi-output) radio design that takes advantage of multipath to actually boost throughput and reliability. After decades of research MIMO technology will now be widely deployed in 802.11, 802.16 (WiMAX) and cellular networks. In addition, dense deployment of lightweight APs (& controllers) will improve stability through load balancing and dynamic radio management. Co-channel interference can be reduced (but not eliminated) by using 5 GHz spectrum and modern Wireless LAN systems (controller + APs). In addition, 802.11k (radio resource management) and 802.11v (station control) will be ratified by the IEEE in the next 12 - 24 months. Taken together these two new standards will make a WLAN more like the cellular network (where the network controls roaming - rather than the station) and will greatly increase wireless stability.