Re: 10 Mbit/s problem in your network
On 9 February 2013 22:49, Keith Medcalf <kmedcalf@dessus.com> wrote:
Most of these networks are provided by "Internet Marketing Companies". In exchange for free-reign in data harvesting and data capture/logging/tracking and advertisement/javascript insertion in web pages (etc), the hotel gets to offer "free" internet connections. Often the Hotel Internet is a profit center for the Hotel, the "Internet Company" paying the Hotel for unrestricted diddling rights to the unsuspecting guests traffic.
Same applies to almost every business that offers "free complimentary internet connections" ...
Occasionally you run into a Hotel that offers a quality and clean internet connection, however, these are few and far between ...
Several 2.5* / 3* hotel managers I spoke with volunteered, implied or confirmed that they're paying on the order of 2k$/mo for "internet" in Northern California. And at least in the US, I'm yet to encounter a complementary WiFi at any hotel that would be doing JavaScript insertion, so I'm not sure where you get your information that the free internet always means ads or a very high level of tampering. One of my prior residential ISPs, Embarq, arguably did more tampering and data mining with my connection than any of the hotels I have ever stayed at. (I'm talking about DNS hijacking.) Now. Notice that these hotels are already paying 2k$/mo and getting 10Mbps, which residentially retails at 40$/mo. How much will 100Mbps cost them? What, still 2k/mo? What are they waiting for? Or, pardon my residential bias, but are some of them still using T1's? Don't those cost a fortune? Wouldn't they actually save their money by going elsewhere? I hear microwave links are pretty popular these days, and offer great bandwidth and latency. C.
-----Original Message----- From: Mike Lyon [mailto:mike.lyon@gmail.com] Sent: Saturday, 09 February, 2013 23:23 To: Constantine A. Murenin Cc: North American Network Operators' Group Subject: Re: 10 Mbit/s problem in your network
"why do the sub-contracted internet support companies design and support such broken-by-design setups?"
Because they don't know any better and lack the technical clue on how to implement a network that can support a hotel-full (or half-full) of people...
But i'm sure they all have their MCSEs and CCNAs so they are qualified :)
-mike
Sent from my iPhone
On Feb 9, 2013, at 19:57, "Constantine A. Murenin" <mureninc@gmail.com> wrote:
why do the sub-contracted internet support companies design and support such broken-by-design setups?
On Sat, Feb 16, 2013 at 6:12 PM, Constantine A. Murenin <mureninc@gmail.com>wrote:
And at least in the US, I'm yet to encounter a complementary WiFi at
any hotel that would be doing JavaScript insertion, so I'm not sure
where you get your information that the free internet always means ads or a very high level of tampering.
They exist, although they are rare. eg, http://bits.blogs.nytimes.com/2012/04/06/courtyard-marriott-wifi/ (This particular hotel apparently stopped shortly after this news broke) On Sun, Feb 10, 2013 at 8:11 AM, Måns Nilsson <mansaxel@besserwisser.org> wrote:
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero. VPN connections are obviously common, but are becoming fewer and fewer by the day - especially non-split tunnel VPN. An on-site transparent proxy(with or without cache) will improve performance to at least some extent, if only because it's isolating the issues of the local network (potentially congested wifi in an environment that really isn't designed for good wifi coverage!) from the upstream. It's far better (and quicker) to handle a dropped packet between the client and the proxy than between the client and the webserver.
From personal experience (around a dozen different hotels this year already) the best thing you can to do improve performance is to avoid Wifi and revert to a wired connection - or if you really want a wireless connection take your own travel wifi router and connect it via a wired connection. The performance difference in many hotels is significant, showing that the problem is often less the hotels Internet connection, and more their wifi.
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's... Scott.
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A). Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA #natog +1 727 647 1274
On 2/17/13 8:33 AM, Jay Ashworth wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones. The chances of the average hotel wifi user even knowing what SSH means is close to zero. {{citation-needed}} The crapy facebook games that everyone plays are both latency senstive and unhappy when their connections are reset. zynga poker peaked at something like 38 million players (per wikipedia)
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's... Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
Cheers, -- jra
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses. Owen
----- Original Message -----
From: "Owen DeLong" <owen@delong.com>
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
No, my ThinkPad doesn't *do* N, 5GHz or otherwise. Neither does my Sprint EVO, nor, as near as I can tell, the Galaxy S4 I'm going to replace it with this year (though on that one, I'm a tad less certain). I'd forgotten that N was dual band, though, yes. I can't say I've ever needed the extra bandwidth N provides, personally, though certainly the hotels we've been discussing might need more to share around. Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA #natog +1 727 647 1274
On 2/17/13 12:18 PM, Jay Ashworth wrote:
----- Original Message -----
From: "Owen DeLong" <owen@delong.com> I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses. No, my ThinkPad doesn't *do* N, 5GHz or otherwise. Neither does my Sprint EVO, nor, as near as I can tell, the Galaxy S4 I'm going to replace it with this year (though on that one, I'm a tad less certain).
I'd forgotten that N was dual band, though, yes. I can't say I've ever needed the extra bandwidth N provides, personally, though certainly the hotels we've been discussing might need more to share around. entirely orthonal to the frequency band used spatial division multipluxing as used by 802.11n is generally going to increase the SNR.
so what you get out of A/N is: * more non-overlapping bands and therefore a much easier map coloring problem) * greater attentuation, which implies more limited range, but also less interferance. * with N-mimo higher SNR if you have >= 2 antennas All of those things make the 5Ghz band a more attractive alternative for lots of applications. given that it's 5Ghz it also requires more power, which is a problem for cellphones, but not so much for tablets and laptops.
Cheers, -- jra
On Feb 17, 2013, at 4:17 PM, joel jaeggli <joelja@bogus.com> wrote:
On 2/17/13 12:18 PM, Jay Ashworth wrote:
----- Original Message -----
From: "Owen DeLong" <owen@delong.com> I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses. No, my ThinkPad doesn't *do* N, 5GHz or otherwise. Neither does my Sprint EVO, nor, as near as I can tell, the Galaxy S4 I'm going to replace it with this year (though on that one, I'm a tad less certain).
I'd forgotten that N was dual band, though, yes. I can't say I've ever needed the extra bandwidth N provides, personally, though certainly the hotels we've been discussing might need more to share around. entirely orthonal to the frequency band used spatial division multipluxing as used by 802.11n is generally going to increase the SNR.
so what you get out of A/N is:
* more non-overlapping bands and therefore a much easier map coloring problem) * greater attenuation, which implies more limited range, but also less interferance.
Greater attenuation is an oversimplification. 5Ghz penetrates things like stucco and concrete better than 2.4. OTOH, 2.4 gets through trees and moist air better. In dry air and/or a vacuum, they're similar. Neither penetrates humans particularly well, though 5 tends to do slightly better.
* with N-mimo higher SNR if you have >= 2 antennas
All of those things make the 5Ghz band a more attractive alternative for lots of applications. given that it's 5Ghz it also requires more power, which is a problem for cellphones, but not so much for tablets and laptops.
OTOH, with 5Ghz, a high-gain antenna is ½ - ⅛ the size (depending on the type of antenna) the size of a 2.4Ghz which also has advantages in portable applications. Owen
On Feb 17, 2013, at 4:32 PM, Owen DeLong <owen@delong.com> wrote:
On Feb 17, 2013, at 4:17 PM, joel jaeggli <joelja@bogus.com> wrote:
On 2/17/13 12:18 PM, Jay Ashworth wrote:
----- Original Message -----
From: "Owen DeLong" <owen@delong.com> I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses. No, my ThinkPad doesn't *do* N, 5GHz or otherwise. Neither does my Sprint EVO, nor, as near as I can tell, the Galaxy S4 I'm going to replace it with this year (though on that one, I'm a tad less certain).
I'd forgotten that N was dual band, though, yes. I can't say I've ever needed the extra bandwidth N provides, personally, though certainly the hotels we've been discussing might need more to share around. entirely orthonal to the frequency band used spatial division multipluxing as used by 802.11n is generally going to increase the SNR.
so what you get out of A/N is:
* more non-overlapping bands and therefore a much easier map coloring problem) * greater attenuation, which implies more limited range, but also less interferance.
Greater attenuation is an oversimplification. 5Ghz penetrates things like stucco and concrete better than 2.4. OTOH, 2.4 gets through trees and moist air better. In dry air and/or a vacuum, they're similar. Neither penetrates humans particularly well, though 5 tends to do slightly better.
* with N-mimo higher SNR if you have >= 2 antennas
All of those things make the 5Ghz band a more attractive alternative for lots of applications. given that it's 5Ghz it also requires more power, which is a problem for cellphones, but not so much for tablets and laptops.
OTOH, with 5Ghz, a high-gain antenna is ½ - ⅛ the size (depending on the type of antenna) the size of a 2.4Ghz which also has advantages in portable applications.
Sorry… Hit send prematurely… An important consideration: A good high-gain antenna helps you with transmit _AND_ receive. More power helps you with transmit.
Owen
On Sun, 17 Feb 2013, Owen DeLong wrote:
Greater attenuation is an oversimplification. 5Ghz penetrates things like stucco and concrete better than 2.4. OTOH, 2.4 gets through trees
My empirical experience with 5GHz says it penetrates concrete a lot less than 2.4. For instance, in one building I was in, 5GHz didn't penetrate the floor so it was only available on the same floor as the AP, but 2.4 GHz worked well both on the floor above and below the AP. This was in a building with quite thick concrete floor, a 3 story town house with the AP placed on the middle. In my current apartment, I moved my AP out of the clothes closet (fairly thin "light concrete" (don't know what it's called) and put it on the wall in my hallway, this increased performance on 5GHz substantially. So I'd like to know where you got your information from because I'd like to read up more because my experience says exactly the opposite. Apart from that, 5GHz is great. More bandwidth, less crowded. -- Mikael Abrahamsson email: swmike@swm.pp.se
On Feb 17, 2013, at 21:12 , Mikael Abrahamsson <swmike@swm.pp.se> wrote:
On Sun, 17 Feb 2013, Owen DeLong wrote:
Greater attenuation is an oversimplification. 5Ghz penetrates things like stucco and concrete better than 2.4. OTOH, 2.4 gets through trees
My empirical experience with 5GHz says it penetrates concrete a lot less than 2.4. For instance, in one building I was in, 5GHz didn't penetrate the floor so it was only available on the same floor as the AP, but 2.4 GHz worked well both on the floor above and below the AP. This was in a building with quite thick concrete floor, a 3 story town house with the AP placed on the middle.
The floor isn't just concrete. Many industrial floors include solid steel plating in the floor. 5 Ghz will not penetrate that and neither will 2.4 (at least not very well). A town house is also likely to have some form of metal plating (or at least a very high concentration of rebar) in the concrete between floors as well, so, I suspect your issue was the metal, not the concrete. 2.4Ghz probably found a path around the outside of the building and back in. 5Ghz once it starts in a direction tends to continue in that direction. It doesn't bounce or curve well at all. 2.4Ghz tends to do better at that, creating the illusion of lesser attenuation. As I said, attenuation is an oversimplification. RF path identification and multipath can get very complex very quickly.
In my current apartment, I moved my AP out of the clothes closet (fairly thin "light concrete" (don't know what it's called) and put it on the wall in my hallway, this increased performance on 5GHz substantially.
So I'd like to know where you got your information from because I'd like to read up more because my experience says exactly the opposite.
Without knowing the details of the makeup of the walls in your closet, I have to say that seems a bit odd to me. Perhaps there is another explanation. The reason 5Ghz penetrates stucco better, for example is that the 23cm wavelength is more than 4x the size of the openings in most of the chicken wire used to adhere stucco to walls. The 12cm wavelength of 5Ghz, OTOH, goes through quite nicely. Owen
On Mon, 18 Feb 2013, Owen DeLong wrote:
The reason 5Ghz penetrates stucco better, for example is that the 23cm wavelength is more than 4x the size of the openings in most of the chicken wire used to adhere stucco to walls. The 12cm wavelength of 5Ghz, OTOH, goes through quite nicely.
<http://www.ko4bb.com/Manuals/05)_GPS_Timing/E10589_Propagation_Losses_2_and_5GHz.pdf> "Aside from large cement blocks and red bricks that displayed somewhat more loss at 5 GHz than at 2.4 GHz (Table 3), losses for all other materials tested were very much the same in both frequency regimes." Looking at their chart on page 9, I see substantially higher attenuation for cinder blocks, 5% lower attenuation with stucco, and 3x attenuation through red bricks. If concrete (10cm think or even more) is similar to red bricks in attenuation, then that would explain the behaviour I have observed in real life. The only material that 5GHz had a lot lower attenuation with was diamond mesh. -- Mikael Abrahamsson email: swmike@swm.pp.se
On 2/18/13 1:42 AM, Owen DeLong wrote:
On Feb 17, 2013, at 21:12 , Mikael Abrahamsson <swmike@swm.pp.se> wrote:
On Sun, 17 Feb 2013, Owen DeLong wrote:
Greater attenuation is an oversimplification.
Along some dimensions sure, e.g. we have quite a lot of parameters we can fiddle with. With respect to an istropic raditor and the same power level it is not. It's about 6-7dB depending on which end of the bands we're comparing - e.g. friis trasmission equation.
On Feb 18, 2013, at 3:07 PM, joel jaeggli <joelja@bogus.com> wrote:
On 2/18/13 1:42 AM, Owen DeLong wrote:
On Feb 17, 2013, at 21:12 , Mikael Abrahamsson <swmike@swm.pp.se> wrote:
On Sun, 17 Feb 2013, Owen DeLong wrote:
Greater attenuation is an oversimplification.
Along some dimensions sure, e.g. we have quite a lot of parameters we can fiddle with.
With respect to an istropic raditor and the same power level it is not. It's about 6-7dB depending on which end of the bands we're comparing - e.g. friis trasmission equation.
Show me a wifi access point for 802.11n that uses an isotropic radiator and I'll consider that more relevant. (Yes, I'm aware that an isotropic radiator is useful as a baseline comparison because it eliminates antenna issues, near-field/far field issues, and a host of other complications. However, the purpose of an isotropic radiator is, at its core, the very definition of oversimplification because it is a theoretical antenna which removes all of the real world complexities. To the best of my knowledge, nobody has ever actually built an isotropic radiator, though there are a couple of very complex antennas that come a little closer than a ¼ wave whip.) Owen
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary. Frank -----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses. Owen
Correct. However, while A is 5Ghz (only), it's not significantly better than G. The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better. Owen On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
I should probably know this, but doesn't N just spread better and have the ability to send receive on multiple polarizations? As an RF engineer I should probably know this, but I can't think of many people in my industry who really care about 802.11_. I really don't even use wireless in my house, though it's generally due to overcrowding the spectrum in populous areas.
From my Android phone on T-Mobile. The first nationwide 4G network.
-------- Original message -------- From: Owen DeLong <owen@delong.com> Date: 02/25/2013 8:38 AM (GMT-08:00) To: Frank Bulk <frnkblk@iname.com> Cc: NANOG <nanog@nanog.org> Subject: Re: 10 Mbit/s problem in your network Correct. However, while A is 5Ghz (only), it's not significantly better than G. The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better. Owen On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
N has a number of advantages… Better spread, the ability to take advantage of polarization, better use of MIMO, and IIRC, a better encoding scheme that allows denser constellation points (more bits per signaling element). N on 5Ghz takes advantage of the increased bandwidth of the 5Ghz channel where A merely replicated G on 5Ghz for all practical purposes. Owen On Feb 25, 2013, at 8:42 AM, Warren Bailey <wbailey@satelliteintelligencegroup.com> wrote:
I should probably know this, but doesn't N just spread better and have the ability to send receive on multiple polarizations? As an RF engineer I should probably know this, but I can't think of many people in my industry who really care about 802.11_. I really don't even use wireless in my house, though it's generally due to overcrowding the spectrum in populous areas.
From my Android phone on T-Mobile. The first nationwide 4G network.
-------- Original message -------- From: Owen DeLong <owen@delong.com> Date: 02/25/2013 8:38 AM (GMT-08:00) To: Frank Bulk <frnkblk@iname.com> Cc: NANOG <nanog@nanog.org> Subject: Re: 10 Mbit/s problem in your network
Correct. However, while A is 5Ghz (only), it's not significantly better than G.
The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better.
Owen
On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
If you want to see something pretty amazing, check this out.. http://www.popsci.com/science/article/2012-06/twisting-signals-vortex-resear... These guys got close to 100 bits/hz using Orbital Angular Momentum in addition to the normal Spin Angular Momentum. There is a picture out there of the I/Q showing the constellation, which to me looks like the future of communications systems. In my world, if you could offer 5 bits/hz or higher you would very likely be able to retire on your own island. Space segment for satellite systems can cost as much as 175k for 36MHz, so giving someone a 20x bandwidth increase would be an absolute game changer. Don't be surprised if you see the 802.11 guys trying to figure out how to make OAM work, it would essentially solve the worlds bandwidth problems at nearly all frequencies. From: Owen DeLong <owen@delong.com<mailto:owen@delong.com>> Date: Mon, 25 Feb 2013 08:56:05 -0800 To: User <wbailey@satelliteintelligencegroup.com<mailto:wbailey@satelliteintelligencegroup.com>> Cc: Frank Bulk <frnkblk@iname.com<mailto:frnkblk@iname.com>>, NANOG <nanog@nanog.org<mailto:nanog@nanog.org>> Subject: Re: 10 Mbit/s problem in your network N has a number of advantages… Better spread, the ability to take advantage of polarization, better use of MIMO, and IIRC, a better encoding scheme that allows denser constellation points (more bits per signaling element). N on 5Ghz takes advantage of the increased bandwidth of the 5Ghz channel where A merely replicated G on 5Ghz for all practical purposes. Owen On Feb 25, 2013, at 8:42 AM, Warren Bailey <wbailey@satelliteintelligencegroup.com<mailto:wbailey@satelliteintelligencegroup.com>> wrote: I should probably know this, but doesn't N just spread better and have the ability to send receive on multiple polarizations? As an RF engineer I should probably know this, but I can't think of many people in my industry who really care about 802.11_. I really don't even use wireless in my house, though it's generally due to overcrowding the spectrum in populous areas.
From my Android phone on T-Mobile. The first nationwide 4G network.
-------- Original message -------- From: Owen DeLong <owen@delong.com<mailto:owen@delong.com>> Date: 02/25/2013 8:38 AM (GMT-08:00) To: Frank Bulk <frnkblk@iname.com<mailto:frnkblk@iname.com>> Cc: NANOG <nanog@nanog.org<mailto:nanog@nanog.org>> Subject: Re: 10 Mbit/s problem in your network Correct. However, while A is 5Ghz (only), it's not significantly better than G. The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better. Owen On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com<mailto:frnkblk@iname.com>> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com<mailto:jra@baylink.com>> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au<mailto:scott@doc.net.au>>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
Owen DeLong <owen@delong.com> writes:
N on 5Ghz takes advantage of the increased bandwidth of the 5Ghz channel where A merely replicated G on 5Ghz for all practical purposes.
You have that backwards, actually, but the legacy support in 802.11g for 802.11b clients does represent a performance hit even in the absence of b-only clients, so claiming that a and g are equivalent is only true on paper. -r (802.11a user before 802.11g, still love the relatively unoccupied 5 ghz spectrum)
Perhaps I don't understand.. Generally in wireless we look at two things; bits to hertz and noise components. If the noise is LESS and the carrier is the same power spectral density, you will have a greater c/n. I've always wondered why wifi didn't implement an array of modcods which can be used with a given system. That way, when you attenuate you have lower efficiency modulation and coding which will allow you to deal with fades better. Maybe they do us it and I'm just not hip to 802.11?
From my Android phone on T-Mobile. The first nationwide 4G network.
-------- Original message -------- From: Rob Seastrom <rs@seastrom.com> Date: 02/26/2013 3:40 AM (GMT-08:00) To: Owen DeLong <owen@delong.com> Cc: Warren Bailey <wbailey@satelliteintelligencegroup.com>,NANOG <nanog@nanog.org> Subject: Re: 10 Mbit/s problem in your network Owen DeLong <owen@delong.com> writes:
N on 5Ghz takes advantage of the increased bandwidth of the 5Ghz channel where A merely replicated G on 5Ghz for all practical purposes.
You have that backwards, actually, but the legacy support in 802.11g for 802.11b clients does represent a performance hit even in the absence of b-only clients, so claiming that a and g are equivalent is only true on paper. -r (802.11a user before 802.11g, still love the relatively unoccupied 5 ghz spectrum)
On 26/02/13 17:19, Warren Bailey wrote:
Perhaps I don't understand.. Generally in wireless we look at two things; bits to hertz and noise components. If the noise is LESS and the carrier is the same power spectral density, you will have a greater c/n. I've always wondered why wifi didn't implement an array of modcods which can be used with a given system. That way, when you attenuate you have lower efficiency modulation and coding which will allow you to deal with fades better. Maybe they do us it and I'm just not hip to 802.11?
They do it, all right, and much, much more, including MIMO -- 802.11 has evolved into something only marginally less complex than the mobile phone wireless stack in the process. -- N.
On 2/25/13 8:42 AM, Warren Bailey wrote:
I should probably know this, but doesn't N just spread better and have the ability to send receive on multiple polarizations? That would be a rather extreme over-simplifcation of spatial-division-multiplexing and space-time-coding. As an RF engineer I should probably know this, but I can't think of many people in my industry who really care about 802.11_. I really don't even use wireless in my house, though it's generally due to overcrowding the spectrum in populous areas.
From my Android phone on T-Mobile. The first nationwide 4G network.
-------- Original message -------- From: Owen DeLong <owen@delong.com> Date: 02/25/2013 8:38 AM (GMT-08:00) To: Frank Bulk <frnkblk@iname.com> Cc: NANOG <nanog@nanog.org> Subject: Re: 10 Mbit/s problem in your network
Correct. However, while A is 5Ghz (only), it's not significantly better than G.
The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better.
Owen
On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones. The chances of the average hotel wifi user even knowing what SSH means is close to zero. {{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's... Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
There's only 83.5 MHz to work with at 2.4 GHz, while in most countries you have at least two hundred MHz in the 5 GHz range (http://en.wikipedia.org/wiki/U-NII). So if you choose to have 40 MHz channels for increased throughput, you can have many more (non-overlapping ones) at 5 GHz than 2.4 GHz, increasing Mbps/area. Frank -----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Monday, February 25, 2013 10:34 AM To: Frank Bulk Cc: NANOG Subject: Re: 10 Mbit/s problem in your network Correct. However, while A is 5Ghz (only), it's not significantly better than G. The true performance gains come from 5Ghz and N together. N on 2.4Ghz has limited benefit over G. N on 5Ghz is significantly better. Owen On Feb 24, 2013, at 8:56 PM, "Frank Bulk" <frnkblk@iname.com> wrote:
The IEEE 802.11n standards do not require 5 GHz support. It's typical, but not necessary.
Frank
-----Original Message----- From: Owen DeLong [mailto:owen@delong.com] Sent: Sunday, February 17, 2013 2:07 PM To: Jay Ashworth Cc: NANOG Subject: Re: 10 Mbit/s problem in your network
On Feb 17, 2013, at 08:33 , Jay Ashworth <jra@baylink.com> wrote:
----- Original Message -----
From: "Scott Howard" <scott@doc.net.au>
A VPN or SSH session (which is what most hotel guests traveling for work will do) won't cache at all well, so this is a very bad idea. Might improve some things, but not the really important ones.
The chances of the average hotel wifi user even knowing what SSH means is close to zero.
{{citation-needed}}
As an aside, I was sitting in JFK airport (terminal 4) a few days ago and having a shocking time getting a good internet connection - even from my own Mifi. I fired up inSSIDer, and within a few seconds it had detected 122 AP's...
Yup; B/G/N congestion is a real problem. Nice that the latest generation of both mifi's and cellphones all seem to do A as well, in addition to current-gen business laptops (my x61 is almost 5 years old, and speaks A).
I think by A you actually mean 5Ghz N. A doesn't do much better than G, though you still have the advantage of wider channels and less frequency congestion with other uses.
Owen
participants (11)
-
Constantine A. Murenin -
Frank Bulk -
Frank Bulk (iname.com)
-
Jay Ashworth -
joel jaeggli -
Mikael Abrahamsson -
Neil Harris -
Owen DeLong -
Rob Seastrom -
Scott Howard -
Warren Bailey