Larry Press wrote:
Have any of you folks looked at the WiMAX mesh specification? If so, what
do you think.
I think it seems reasonable for a commercial deployment, but the tree structure and TDMA things they advocate might be a bit too rigid for a more anarchic consumer-owned network. However, I never saw the mesh spec, only some presentation on the topic, from Nokia.
The general feeling I get from the 802.16 spec is that _everything_ is in there. To me, that has a bad smell to it, and I've heard comments drawing a parallel to wireless ATM, which had similar characteristics, and flopped miserably.
The only real problem(s) with ATM were first the "cell tax". (5 out of every 53 bytes are ATM "overhead") and second increasing bandwidth of Ethernet eclipsed the QoS architecture of ATM. With gigabit Ethernet now a proven technology with declining prices, and 10 gigabit Ethernet just around the corner, it is increasingly difficult to argue that bandwidth is scarce enough within an enterprise to justify complex bandwidth management techniques, such as ATM.
The last bastion of ATM wass in the wide-area, and there are several reasons for this. First is the predisposition of telcos to use a circuit-oriented technology, second is the legitimate requirement of telcos to be able to multiplex other layer 2 services (e.g. SMDS, Frame Relay, Transparent LAN Service, etc) over their networks, and finally, even within a pure IP environment, there is value in having a mechanism for creating virtual circuits. As Internet packet traffic absorbs everything else, issue two evaporates, though some form of virtual circuit capability will continue to be useful for QOS and routing management.
Finally advances in optical communications such as DWDM (Dense Wave Division Multiplexing) have drastically reduced the cost of wide-area bandwidth relative to what it was when ATM was invendted, thus reducing the need to micro-manage small chunks of bandwidth via traditional virtual circuits and admission controls (the data equivalent of busy signals). Today, nearly every large ISP has opted for packet-over-SONET backbones rather than national ATM clouds.
I see two major "flaws" with 802.16. The first is that the DOCSIS-based MAC is overblown for what is needed from 802.16, but this is what happens sometimes when you design protcols with committees.
The second, and more damning is that .16 has been hyped (mostly by Intel) to provide reliable fixed wireless services, potentially over unlicensed spectrum. This is impossible once the interference environment becomes "uncontrolled". Any nearby interferer (and potentially any distant interferer as well) can make your base station unable to decode a distant client.
But wait, we're back to that again. Sorry!
The huge success of 802.11 might prove a big hurdle to its younger brother, which apparently feels it has something to prove. ;-)
The Intel hype machine rides again.
But yes, the idea behind 'WiFiMaX' (and projects like it) is that cheap OFDM PHYs that will run in 2.4GHz and 5GHz are here today in the form of 802.11a and 802.11g cards. Given the lack of firmware in modern designs, we can make the MAC do mostly what we want it to do, rather than suffer with the 802.11 MAC which is entirely sub-optimal once it is no longer tasked with providing LAN-like services over short distances.
Except for very specific situations, I prefer to mesh at layer 3 with OLSR or TBRPF.
Interesting. Among the mesh networking products out there, there seems to be a divided camp.
> Would you care to share what you feel the relative advantages of layer > 3 meshing are?
For in-office (or in-home) scenerios, layer 2 meshing is fine, thought its often better to pull a wire to the next AP. Still, by connecting things at layer 2, all of the autoconfiguration stuff (DHCP, router discovery, neighbor discovery (ARP in IPV4), service discovery (mDNS, SLP, JINI, UPnP) just works.
That IP's services and archicture provide scale. Just ask anyone who has ever attempted to buid a large-scale bridged network.
Then there are the "layer 2.5" designs, but these are just routing (though not IP routing). Why re-invent the wheel? I openly admit that I am an IP bigot.
Jim
