[ NNSquad ] Wired vs. Wireless performance not a good debate to have

["David P. Reed" <dpreed@reed.com>]

  [ This one slips into the window of the thread closure, but more on
    the wired vs. wireless topic will be dependent on actual new
    information of value being imparted in submitted messages.

      -- Lauren Weinstein
         NNSquad Moderator ]


I am currently engaged in a study (along with others, as Dave Farber is 
aware) of wireless access network configuration problems, in particular 
on AT&T Wireless 3G, but the same sort of problems have been observed on 
other 3G access networks.

The performance issues have nothing to do with the radio link, medium 
access control, etc.  The problem is in the routers and switches in the 
portion of the network upstream from the cell sites.   Since the study 
is not complete, I am not in a position to reveal the data, or provide 
final conclusions.

However, I can assure you all that debating the behavior of carrier data 
wireless (WiMAX, HSPA, or CDMA) networks in ignorance of detailed 
configuration information or measurements is not going to shed light on 
"wireless" issues as different from "wired" - the elements of the 
network with problems are all "wired".

The class of problems being analyzed are essentially identical to 
problems that have been seen in wired access networks (such as the 
DOCSIS problems observed in Comcast's network) where the layer 2 service 
is configured to create persistent congested states in underprovisioned 
upstream layer 2 elements.

I hope that the operators who have such configuration issues will be 
able to improve their network architecture with the proper fixes.