802.11n poll - legacy support or not?
Last Post: February 10, 2006:
The current .11n draft board is trying to decide between supporting or not supporting legacy PHYs and MACs in .11n. Here are the alternatives. First, create an entirely new PHY for .11n that allows greater than 100Mbps throughput with AES encryption or higher (future) only. This could be pushed out pretty quickly and wouldn't require chip manufacturers to provide backward compliance with legacy PHYs and MACs (i.e. .11abg and TKIP encryption support). Or, second, wait until the .11n drafts are capable of defining exactly what should be supported and backward compliant in .11n (big controversies surrounding this). This will take a lot of time to accomplish and will likely push out .11n for another two years or so. If the first option is taken, consumers will be given the opportunity to migrate from older, slower systems to much faster newer systems that don't have support for the old products but the standard will be completed probably in less than a year. So, is it more important to support legacy products with .11n with some increased throughput or would it be better to deploy "pure" .11n environments with huge throughput gains and even better future capabilities?
What are your thoughts?
Does 802.11n mean multiple antennas and will there be external antennas on these devices that are detachable?
The reason I ask this is, won't it change the dynamics of the 802.11n PHY if someone tries to place a higher gain antenna on one or more of thee MIMO (802.11n) antenna ports?
Great discussion. I have a few of your co-workers in class here in NC.
I think that they should come out with an entirely new standard. If not, the same thing will happen to wireless networking as happened with the Mustang until last year. Ford used the same chassis for 20+ years, and no matter what kind of body you put on it is still couldn't handle like a real sports car. </bad analogy>
It may be a tough road with all new technology and no backward compatibility, but should be worth it for the long term. My first thought is that initial adoption may be slow, but by the end of this year and in 07 enterprises will be so hungry for more wireless bandwidth that I think they will be chomping at the bit to get more speed. In a year or so I can look in the Planet3 forum archives and see if I'm right. :)
The manufacturers could add backwards compatibility support to their cards, which they would have to do to be competitive, without it being defined in the standard.
Joel has a future in politics, I can tell. If 802.11n without backwards compatibility was the Republican Party, Joel would be Fox News.
Zwwwiiiinnnnngg.... that was that sound of that comment going right over my head. Not being politically oriented even one tiny particle, I have no idea what you meant by it. I don't watch Fox news (or any news really, except when I'm sitting in the airport)... I watch TiVo. For the last three elections, the write-in candidate that received my vote was "None of the above" (seriously) because I disliked the candidates so much. Please explain how you felt my questions were skewed.
My intent of this poll was to get an idea of how people felt about going forward with .11n technology because I am involved in many discussions concerning it (and our take on it) inside Cisco. I honestly wanted to know what my peers thought so that, if the opportunity arose, I could express that opinion to my Cisco peers. Were the poll questions biased?
I don't think your question was biased, but the question itself is a political one for sure. I know exactly what Ben was getting at - and I thought the same thing upon reading the question. Here is my $0.02.
802.11g blows chunks - why? Simple - protection mechanisms. Why do we use protection mechanisms? Simple - because old cards only understand their modulation while new cards understand both old and new modulations (DSSS and OFDM).
Think about what will happen if you REQUIRE .11n cards to use protection....it will turn into .11g all over again. .11a ruleZ because it doesn't use protection. What about the upgrade/migration process you ask? Glad you asked - here's my OTHER $0.02.
In the enterprise, if an organization already has .11a/g rolled out, then that won't just magically go away - too large of an investment. If it's working, then they'll stick with it. If they have to have more bandwidth (a good reason for upgrading to .11n), then they'll roll out .11n APs where they're needed only. They'll be co-located with whatever else is there already. Will it be a perfect environment? Obviously not. There will be some interference, but considering the problems .11g already has of losing 40% of its throughput due to using protection - who's going to notice? Eventually, the proliferating need for more bandwidth and the availability of new .11n equipment will cause anything that is competing with .11n equipment to be removed slowly.
This is a scenario exactly like having .11b and .11g co-located where the .11g system is configured for ONLY OFDM data rates (we call this Pure-G). There is some interference between .11b and .11g, but it only causes a significant problem when both systems are moving lots of data. I think the .11n TG should NOT make the same mistake the .11g TG made with backwards compatibility. Remember my whitepaper on Protection Ripple? We really don't want that scenario again.
We want to give organizations an upgrade path to higher throughput, w/o having to say, "but if .11b/g is present, your .11n gear really doesn't give you much advantage....sorry that you wasted all that money."
That's my take as well and, now that the poll has been here for awhile, I'm willing to add my two cents as well. I'm dead on in agreement with what you're saying Devin. I think it should be treated as a new technology, something to migrate to if that much throughput is needed, not something that once again tries to please everyone by being completely backward compliant. Maybe we should start using the term "immigrate" rather than "migrate" since immigrate implies more of a permanent move while migrate means you might be able to go back.
Sometimes we need to take the next step and move on with things (in life as well as technology). Unfortunately that's not the way most people think. Most people think we should coddle everyone and make sure no one gets left behind and that the dollars spent yesterday should apply to tomorrow's solutions as well. Often this limits the capabilities of new technologies -- just look at all the problems it causes Microsoft. They still support DOS-based programs! I mean come on people, give VisiCalc a rest, why don't cha!?
So, yes, I'm totally for designing .11n with no backwards compliance requirements because it will give manufacturers the opportunity to create a very high-throughput wireless solution without the need to support all the "junk" technology (WEP, CCK, etc.) from .11abg. We can take the best of the best (OFDM, AES, MIMO, etc.) and move forward to the future of wireless networking where it will be possible to achieve datarates as high as 600 Mbps (maybe more).
But that's just my two cents.
Generally Fox News acts as if they were an arm of the Republican Party (not claiming they actually are, just claiming they are commited to the cause). When Fox News gives a "news" report, generally they make sure to sway their less aware viewers in favor of the conservative agenda while still tacitly giving both sides. It helps them draw huge numbers of conservative viewers because most people want their news to be slanted in favor of their own opinions.
It was clear from your post that you are strongly in favor of 802.11n not having backwards compatibility. If you actually want to post an unbiased poll, then post the poll without the comments. By adding your slanted comments to the poll, you are making the results of the poll worthless.
As far as the poll question goes, I favor backwards compatibility. Without it, I think large network owners will be reluctant to upgrade.
I think it is up to vendors to allow Protection to be disabled as a configurable setting on the AP. I noticed that some Linksys wireless routers with relatively new firmware allow Protection to be disabled even in b/g mixed mode. This accomplishes what Devin is advocating without requiring a full network overhaul.
I know the danger here is that IT managers may cause problems on their own network by not knowing when they should disable Protection and when they should enable it. I believe that's what the CWNP Program is for in the first place. If IT managers hire CWNAs, those folks should know when to enable or disable 802.11n Protection (assuming it gets ratified in line with the last EWC draft).