Isn't there something whereby the network runs as fast as the slowest device?
I was wondering if any has stopped supporting old devices e.g the 802.11b range?
I've had a problem with wifi dropping out in the main student common room and it seems to be a couple of ipods and nintendo DS devices that are running on 802.11b range that's bringing everything to a screeching halt. After disabling 'b' on a few AP's nearby the speeds are much better.
We have a Xirrus managed system in place and I've already got it carved up with 1/4 dedicated to 5ghz and the rest on bgn. I'm planning 1/4 5ghz, 2/4 gn and the other 1/4 bgn
I hope that makes sense, we still have huge amounts in the 2.4ghz range
Interested to hear how everyone else carves up their wireless space?
Isn't there something whereby the network runs as fast as the slowest device?
Basically, I understand there is a lot more to it than that but certainly allowing 802.11b affects anyone else connected on that same AP. I'd rather shut down the 'b' sitewide but knowing we do still have students using it that would be a bit short sighted
That was the case on older routers/APs, but with newer ones that's not the case. To be honest I'd set a deadline and close off 802.11b. It only supports up to 11Mbps which isn't a lot in bandwidth terms.
I've checked and the iPod 1st gen has a b/g card. The first generation DS's only support WEP and not WPA, so you're stuck there. Who still uses the 1st gen anyway!?
Assuming that I correctly understand your description of the problem; your speeds are dropping because of b-mode clients.
802.11b is a DSSS technology. (Direct Sequence Spread Spectrum)
802.11g/n is OFDM (Orthogonal Frequency Division Multiplexing)
When your AP is set to mixed mode, it has to have a way for both MCS's (Modulation & Coding Scheme) to exist on the same BSS. (Basic Service Set)
That's done through what's called Protected Mode. It's because 802.11 networks are still only half-duplex and they use CSMA/CA to avoid collisions. I freely admit to being too lazy to get into a whole discourse on things like Network Allocation Vectors, but the basic deal is that the moment a mixed mode AP hears a probe request from a DSSS or HR-DSSS client, it goes into protected mode. They don't even have to associate with the AP to trigger Protected Mode. They only have to send out a single probe request. Even worse... Your AP doesn't even have to hear that "ping" from an authorized client. It can be ANY client station that happens to be within range.
On a 54meg connection that would otherwise have an aggregate throughput of 20-25meg, it will actually drop to less than 10meg aggregate.
The solution is to either have co-located WAP's using UTP cable for back-haul to the network, with "b" clients segregated on one and "g/n" clients on the other; or turn off mixed mode altogether.
I don't take it personally, so there is no need for you to fret about that. But if you believe that my advice is incorrect and believe it so strongly that you feel it necessary to comment upon it, then it is incumbent upon you to explain why.
And for what it's worth, I'm pretty sure you don't have an obligation to care whether I understand you, although I do appreciate the concern for my feelings.
But let me see if I can show my gratitude by helping to narrow my ignorance down for you a bit so as to save you some effort in educating me.
802.11b certainly is Direct Sequence Spread Spectrum, specifically HR-DSSS using Complimentary Key Coding for modulation as opposed to DSSS and Barker Code as defined under 802.11 Prime. So you couldn't be talking about that...
802.11g specifies ERP-OFDM along with ERP-DSSS/CCK for backward compatibility with 802.11b and specifies mandatory data rates of 6, 12 and 24 in "G" mode and 1, 2, 5.5 & 11 in "B" or "Mixed" mode. So you couldn't be talking about that either...
802.11n or HT-OFDM, leaving aside the differences in bandwidth (20Mhz vs 40MHz) is not really any different from "G" in it's behavior. In fact, the 802.11n Amendment only specifies 3 mandatory technologies:
- Support for at least two spatial streams on the AP.
- Aggregate Media Access Control Protocol Data Units in transmit mode and Aggregate Media Access Control Service Data Units in receive mode
- Block ACK
So... you couldn't be talking about that either.
So what's left...
Oh yes... Protected mode.
When running an AP in B/G/N mode or "Mixed Mode", 802.11-2007 requires backwards compatibility for the client stations. The data rates are defined by a Modulation and Coding Scheme Matrix so the AP can talk to client stations based on the client's ability. The problem is that new developments in Full Duplex WiFi using "noise" cancelling techniques notwithstanding, WiFi is a half-duplex medium that used CSMA/CA or Carrier Sense Multiple Access / Collision Avoidance. Unfortunately, the way that it avoids collisions is through a Network Allocation Vector or NAV Timer. A station can't broadcast until it's NAV timer is zero. When a data frame is transmitted the Duration/ID field is used by the other stations to reset their NAV timers and begin the count to zero again. But "B" stations won't understand "G/N" station frames because DSSS and OFDM are not mutually understandable. So, in order to compensate the "G/N" stations begin to use RTS/CTS (Request to Send / Clear to Send) or CTS-to-Self using a Modulation method which the "B" stations DO understand. The problem is that this adds so much overhead that whereas you can usually count on getting a throughput of about half of the claimed data rate, protected mode cuts it in half again. Which certainly SEEMED to be the behavior described in the original post. So it couldn't be that either...
Perhaps you were referring to my failure to discuss Packet Binary Convolutional Code, Differential Binary Phase Shift Keying, Differential Quadrature Phase Shift Keying, Binary Phase Shift Keying, Quadrature Phase shift keying or the various levels of Quadrature Amplitude Modulation; OR how the relate to the various available data rates. But then, if I didn't mention it then I couldn't have been wrong about it...
Maybe, you objected to my solution...
But no... co-located WAP's with UTP back-haul (as opposed to Mesh, which loses half it's throughput on every hop.) is certainly a valid solution, as is turning off "B" mode altogether and having a pure OFDM WLAN that wouldn't need a protection system.
Oh wait... I've got it!
You meant that I misunderstood the QUESTION posed by the original poster.
Well, that is certainly possible. If you would be kind enough to explain it to me so that I may learn from my mistake, I would be grateful. Thank you in advance.
And so began the wifi wars.
I thought you would say something like that rather than risk getting into an actual technical discussion.
But I figured, hey... it's your foot, your mouth and your decision about which direction said foot moves.
By the way... Disabling the LOWEST rates (1 & 2 meg 802.11 Prime) improves throughput for the same reason that Airtime Fairness was developed. They are Media hogs, taking up to TEN TIMES longer to transmit the same data as even 802.11b.
Disabling the highest rates reduces you maximum potential throughput.
I'll tell you what. I'm going to let my posts stand, you can do what you like with your posts and everyone who reads the thread can decide for themselves what they think.
I'm going to unsubscribe now and you should feel free to have the last word.
Well, it's irrelevant because OP does not have any "B" only devices, so protection is not the issue.
I am happy to get into the technical side any time you want, but you are wrong on so many levels I wouldn't be able to easily answer you. I would need to run a couple of days of classes to even scratch the surface.
FWIW, disabling "B" rates is a critical mistake. If you want to stop "B" devices from associating, then make 6Mbps a mandatory rate.
Honestly, it's nothing personal, you just don't understand wifi well enough, though you are clearly making an effort to learn. You shouldn't take offence, but the harsh reality is that you are giving out BAD advice from an untrained viewpoint. Your advice is manifestly wrong, I am afraid to say, and I call you out not to embarrass you, but to prevent the OP from making a bad situation worse. Until you know better, please don't give out this kind of advice.
Keep going, by my estimation it will take you about 2 years of full time WiFi design and analysis to truly understand how these things play out in real networks. That's not a reflection on you - it took me 6 years to get my CWNE, this is working as a full time WiFi engineer and trainer. I also write courseware for AirMagnet and teach advanced analysis classes. What you're referring to is probably the most difficult things to troubleshoot and diagnose, many many senior engineers have problems understanding the big picture.
White_Fi (19th March 2014)
You don't have any devices that are capable of ONLY 802.11b, so protection is not likely to be the issue. (There are circumstances where protection may be triggered - neighbour AP's, rogue wifi enabled printers, etc, however I would think unlikely here).
What you described can have many possible causes.
If it's a serious problem you need to have someone run spectrum analysis and a frame capture and look for:
1) Layer 1 problems - non wifi interference, adjacent channel overlap, transmit power
2) Layer 2 problems - co-channel interference, neighbour access points.
To rule out protection, look for a beacon frame from your AP and see if the USE_PROTECTION bit is set to 1 and if the NON_ERP_PRESENT bit is also set to 1.
Look specifically for the device data rates in use compared to the expected rates for the given signal to noise ratio. Get a value for retry rates and CRC errors. Capture the conversation of a specific problematic device, and if possible look for the point that triggers the device to try to roam.
Only then can you begin to think about what to do to put things right.
If I had to guess I would say that you probably have too much power in the system. I don't know how many access points you have, but one starting point would be to reduce the transmit power.
I would also be curious to see how you have the AP's deployed.
One point, on you use of frequencies, the access points should automatically steer devices on to 5 Ghz that are dual band capable.
You should plan your deployment from 5GHz perspective, then only activate 2.4GHz on the access points you need to have on to support the number of 2.4 GHz devices you have.
Again, if this is an important network, and you are having these kind of problems, then pay to get a certified WiFi guy with the correct tools to make an analysis. He should be able to narrow down the problem in half a day if he is any good.
Before engaging someone, ask them for their wifi certifications/qualifications and what tools they use. Ask to see examples of previous jobs.
I hope this helps you to get a solution.
Last edited by neilmac; 19th March 2014 at 10:01 AM.
After extensive testing (using basic tools like Ping / file transfers) and my MiSPY I can say that the best results with WiFi are that you set N only with 20Mhz only. All other settings should be left at default as they're only useful if you are in a terribly congested WiFi area.
(40Mhz is just awful in anything less than a perfect environment)
this is all 2.4Ghz (I don't have 5Ghz yet)
You don't get 40 MHz in 2.4 GHz. Tell me what device you were testing with ?
There's a lot of misconceptions about what 802.11n offers, in 2.4GHz it's all about reliability.
Find me any client device that's capable of 40 Mhz in 2.4 GHz. Tell me what devices you used in your test and I will reply with an explanation.
There are currently 1 users browsing this thread. (0 members and 1 guests)