Bad idea to run mixed 11g/11n...?

[drlandau]

I'm a bit confused, because it seems that most of the reviews I have read says, that in mixed 11g/11n environments, performance will degrade - some even more so than if you just run strictly 11g.

I have a SqueezeBox mediastreamer (11g), a laptop (11g) and a pc that I'm going to buy a new USB adapter for anyway (thinking 11n).

I want to secure the network with at least WPA, possible WPA2.

And I need a strong and steady signal with good performance, because I have a lot of interference from other WiFi's in the area,

So my question is: For the best result, should I:
- choose a 3-antenna 11n router and run mixed mode?
- choose a 3-antenna 11n router and set it to run fixed 11g?
- choose a 3-antenna 11g router?

Any suggestions to router and new USB adapter are most welcome.

Thanks a lot.

/Nicolaj

 

[thiggins]

The third antenna in a draft 11n router is receive only and will, in some cases, provide higher thoughput than a 2 antenna router. But draft 11n does not help reduce interference from neighboring networks. And a draft 11n router set to 40 MHz mode will actually make interference worse for your neighbors.

Running a mix of draft 11n and 11b/g clients on the same draft 11n router will reduce speed somewhat for the draft 11n client but reduce the speed of the 11g clients by more than half.

If you have only one draft 11n client, set your draft 11n router to run in 11g mode.

 

[DaveMcLain]

I posted about this in the thread on the DIR-655. I was given one of these by a friend and at first I set it to use B,G and N figuring that this would be best. In my completely G environment this did NOT work well at all. In fact ANY combination that had N turned on didn't work well with G. I was surprised by this and at first I was thinking that if N was on and I had a G client it would make no difference to the G client. This was not true and some of the G clients would barely function at all.

Now I have the router set to B and G and it works fine with good range etc. Hopefully this will get a lot better when N becomes a final standard. If not it's pretty lousy at least in the 2.4Ghz band. Now in the 5Ghz band how well does N interact with A?

 

[thiggins]

Now in the 5Ghz band how well does N interact with A?

Probably the same. But since there is so little 802.11a gear in use (at least in the U.S.) it's not much of an issue.

Perhaps final 802.11n gear will have better interoperability with 11g, but don't count on it. There is an amazing amount of leeway in all of the 802.11 specs that depending on how makers implement drivers and firmware can be the cause of your problems. In other words, your 11g clients could be operating in a way that wasn't a problem before draft 11n, but could be causing your woes when used with a draft 11n router.

That's why manfs allow you to downshift to limit operation to only the older standards. It's a good idea anyway to run your 11b/g clients on a separate WLAN from your draft 11n due to the poor bandwidth sharing of draft 11n APs/routers.

 

[tipstir]

802.11b/g/n most routers won't work right if you don't set all 3 choices and then use 20/40Hz. AGain everything depends how your dwelling and area. Might for now just use 802.11b/g until the N becomes standard. N works but quality link is stronger where it nearer to the router than when it a few floors away. Using another N access point make help. N is so tricky..

 

[thiggins]

802.11b/g/n most routers won't work right if you don't set all 3 choices and then use 20/40Hz.

I strongly disagree, Tipstir.

Default for Wi-Fi Certification is 20MHz bandwidth and most users should leave that setting alone. 40MHz bandwidth just causes problems with neighboring WLANs and you lose range.

If a user has only 11b/g gear they should disable N mode. This will lessen chances of legacy devices having problems.

 

[DaveMcLain]

I didn't have any wireless gear back when 802.11G was in draft status but were there similar problems at that time? I do own a few pieces of old .11B equipment and they seem to work fine when used with a .11G AP.

I just sort of wonder if these interoperability problems will make it take a LONG time for the N equipment to really take over or they will gradually fade as the standards become more solidly defined and are no longer "draft".

I really appreciate the effort that you put into this site too by the way Mr Higgins. I've learned a lot and I've enjoyed all of your reviews etc. Good job.

 

[thiggins]

Thanks for the kind words, Dave.

Throughput will always be reduced when running a mixed network because the faster clients have to wait for the slower ones. But the bandwidth allocated to at least 11g clients could be improved. However, AFAIK, there is nothing in the standard to mandate this and there is no incentive for manufacturers to optimize performance for "legacy" applications.

Bottom line, don't hold your breath for improved bandwidth for legacy clients in mixed 11n networks.

 

[zjohnr]

Some additional speculations ...

So my question is: For the best result, should I:
- choose a 3-antenna 11n router and run mixed mode?
- choose a 3-antenna 11n router and set it to run fixed 11g?
- choose a 3-antenna 11g router?

Another possibility, albeit one that should be thought through carefully before attempting ... could be to obtain a dual band, dual radio router. In that case you could run 802.11n on the 5GHz band and 802.11g on the 2.4GHz band.

The potential advantage is that 802.11g and 802.11n would not interact in this case. Depending on what interference sources you are dealing with, that might also be minimized by using the 5GHz band for n.

However, there are also many potential disadvantages.

• Not all 802.11n clients support 5GHz. You may not be able to use the 5GHz band with your 802.11n clients. In this case, this approach would be a non-starter for you.
• The range of 5GHz 802.11n is apparently much less than in the 2.4 band. (At least with the current hardware). Depending on what distances you plan to use 802.11n, this could also be deal-breaker for you.
• Dual radio routers cost more than an equivalent router with only a single radio.
• Setup is possibly more complicated than you'd wish to bother with.

The above are only the potential "gotcha's" which came to mind when I wrote this. There are probably others which didn't occur to me.

Caveat emptor.

I strongly disagree, Tipstir.

Default for Wi-Fi Certification is 20MHz bandwidth and most users should leave that setting alone. 40MHz bandwidth just causes problems with neighboring WLANs and you lose range.

Based on my own recent experience, I think it's also worth pointing out that another problem with turning on 40MHz bandwidth is that your clients may not be able to use it. I'm not sure how many clients actually support 40MHz, but apparently the 802.11n on my early 2008 MacBook does not.

I wouldn't be that surprised if a lot of other clients out there also lack support to utilize 40MHz. After all, it's a potential "cost savings" reduction that your typical consumer probably would never even know was "missing".

I'm also not sure how you could determine this before purchase. I have not seen it mentioned in the (few) specifications I've glanced at.

-irrational john

 

[zjohnr]

Mixed 802.11g & n: When is it a problem?

Bottom line, don't hold your breath for improved bandwidth for legacy clients in mixed 11n networks.

Tim, apologies if you've already covered this earlier and I just overlooked it.

This discussion got me wondering (again) if the performance reduction when running mixed is only a problem when 802.11g and n share the same (2.4 GHz) channel and/or radio.

If you had both protocols active in a dual radio context, say by either using two wireless access points or a dual radio router, would there be problems if, for example, 802.11g was on channel 1 and 802.11n on channel 11?

How about if they were on neighboring channels?

-irrational john

 

[thiggins]

Mixed WLAN means 11b/g and draft 11n on the same radio.

 

[airhome1]

conflicting advice about mixed "g" and "n:

Thank you in advance for your patience but I'm still trying to sort out the issues associated with a mixed environment. In my home I've got multiple "g" devices and now one "n" device. I bought a belkin N+ based on your review and several others. Based on the SNB articles about mixed environments I was going to set it to run in "g" mode only. However I found a post at another site (see below) which suggested that I should keep the router running in full mixed mode. I'd be grateful if you'd comment on the advice which I've posted below.

Thank you in advance.

Simply put, "N" is multiple channel "G". Each individual channel will only yield "G" speeds, so unless the network drivers can packetize data onto multiple channels, the connection will be at "G" speeds. "N" itself is not faster than "G", each channel is the same speed, except than "N" has multiple channels, while "G" allows only one. Where "N" comes into it's own is when you have multiple channels. Under "G", with multiple connections, you will fall down to "B" speeds; while under "N", you will gain "G" speeds.

"N" is probably more important in a place where you have WiFi with multiple connections, since each connection will be in a separate channel. There is no serious "downgrading" involved. "N" allows twelve "G" speed channels simultaneously; while "G" would either multiplex to connections, or more commonly, allow a number of "B" speed channels simultaneously.

If "N" and "G" devices are mixed, the network will negotiate "G" connections first, then if there are available channels, you will obtain multiples of "G" speeds. To get full "N" speed, the ethernet connection will need to be Gigabit, since 100 speed Ethernet does not have enough speed for more than 1 "G" connection. And the computers themselves will have to have very low latency to service the connection.

If given the choice, "G" at a lower cost is more effective than "N" when it comes to one or perhaps two interconnections; while "N" will be better at multiple machines that would otherwise push the system to "B" speeds under "G".

 

[thiggins]

Thank you in advance for your patience but I'm still trying to sort out the issues associated with a mixed environment. In my home I've got multiple "g" devices and now one "n" device. I bought a belkin N+ based on your review and several others. Based on the SNB articles about mixed environments I was going to set it to run in "g" mode only. However I found a post at another site (see below) which suggested that I should keep the router running in full mixed mode. I'd be grateful if you'd comment on the advice which I've posted below.

That post demonstrates such a twisted "undertanding" of 802.11n, that I'm speechless...

 

[zjohnr]

However I found a post at another site (see below) which suggested that I should keep the router running in full mixed mode.

Out of curiosity, could you provide more info about this "other site"? A link to the post you quoted from perhaps?

That post demonstrates such a twisted "understanding" of 802.11n, that I'm speechless...

Tim, it would still be helpful for those us who don't know enough to be shocked into silence if you still could post a slightly longer response. Perhaps point out the major inappropriate twists in the excerpt airhome1 quoted from the other site? Otherwise I'm left to scratch my head and think, "That seems wrong, BWTHDIK?"

-irrational john

 

[thiggins]

John, I have addressed the issue of mixing n and "legacy" clients so many times that I've lost count. And there is always this article:
Add, Don't Replace When Upgrading to 802.11n.

Google the first phrase from the post and you'll find the site.

 

[airhome1]

The quotation about mixed "n" and "g" environments was in discussion about the issue at the following URL:

http://www.ehmac.ca/anything-mac/60756-airport-extreme-n-slowed-down-g-devices.html

I'll stick with Tim Higgins advice. In the time that I've been reading this site, his writing and analysis has always seemed thorough, well thought out and relaible.

Thanks

 

[SpiderJon]

John, I have addressed the issue of mixing n and "legacy" clients so many times that I've lost count. And there is always this article:
Add, Don't Replace When Upgrading to 802.11n.

Excellent article - but now two years old.

In it you say "the current crop of draft 11n consumer routers handle two things quite badly: mixes of draft 11n and 802.11g clients; and any security option other than WPA2."

What's the situation with today's "current crop" of 11n consumer routers?

rgds

Spider

 

[thiggins]

What's the situation with today's "current crop" of 11n consumer routers?

Things haven't changed.

Mixing G and N clients will still results in degraded performance for both. See Before you ask about mixing N and A/B/G clients .

Using WEP or WPA/TKIP will also still limit you to 802.11g speeds (54 Mbps maximum link rate). That is due to the 11N spec.

 

[zjohnr]

Any idea why they restricted the link rate in the spec?

Using WEP or WPA/TKIP will also still limit you to 802.11g speeds (54 Mbps maximum link rate). That is due to the 11N spec.

Yes, I'm always puzzled by that. It seems (to me) like a strange restriction to put into a spec. While I'm all for deprecating those older security protocols, silently restricting the link rate seems a rather obscure approach for doing that. No??"

Am I missing the point of the restriction? Is there another reason that they put that into the 802.11n spec?

-irrational john

 

[thiggins]

More accurately, the 11n spec reads that HT modes (High Throughput, i.e. anything higher than 54 Mbps) may be enabled only when using no encryption or when using WPA2/AES.

As to why, you'd have to ask the people on the IEEE task group.