Does 802.11n Spell the 'End of Ethernet'?

alphadogg writes "Is the advent of the 802.11n wireless standard the 'end of Ethernet'... at least in terms of client access to the LAN? That's the provocative title, and thesis, of a new report in which the author began looking into the question when he heard a growing number of clients asking whether it was time to discontinue wired LAN deployments for connecting clients. Would 11n, the next generation high-throughput Wi-Fi, make the RJ45 connector in the office wall as obsolete as gaslights?"

Comments

Um, no.

[morgan_greywolf]

When the Porcine Aviation Assocation makes WiFi as secure as wired LAN, then we'll see the end of Ethernet. Until then, no.

Re:Um, no.

[Praedon]

I think we will always have wired networks, for the simple fact that as technology progresses, so do the methods of spying and such. It's much easier to eavesdrop on WiFi than it is on a wired network. You would need physical access to the wired network in order to carry out your plans for espionage.

Re:Um, no.

[niceone]

And that's going to happen just after the Porcine Aviation Assocation makes WiFi actually run at the speeds that it says in the headlines.

Re:Um, no.

[cayenne8]

Yep...I was thinking one word when I read this:

SECURITY.

Re:Um, no.

[Jaqenn]

I agree that wired LAN is more secure than WiFi. But can't you do some pretty scary signal reconstruction by reading electromagnetic noise coming off your network cable? It's my understanding that this can be done from X yards away, through walls, whatever.

Yeah, that moves your vulnerability away from the hobbyist tier and into the professional tier, but honestly, which one scares you more?

Guess you could always wrap your cable in tin-foil.

Re:Um, no.

[nine-times]

Right. There are too many reasons to use ethernet, and security is just one of them. Ethernet is also more reliable, and it's still faster. 802.11n is not running as fast as 1Gbps (which is what both my home and work network are running at). Give it a couple years, and we'll probably all be running 10Gbps networks, and though wireless speeds will improve too, I see no reason to believe that they'll ever catch up. Also, wired connections are more reliable, easier to control, etc.

Now, I don't see much reason to string ethernet through people's homes, at least not most of the time. Use WPA, secure each of your computers (password protect them and firewall any services you aren't using, preferably don't use Windows). You'll be fine, and 802.11n is probably way faster than any internet connection you might have.

Re:Um, no.

[arivanov]

Pigs can fly. It is a matter of applying sufficient thrust.

It is a matter of contention ratio.

An average office has a contention ratio of 1:100 for server access and it still works. A WLAN contended to 1:100 will not work. So you have to upgrade your porcine fleet with higher thrust engines. You do that by rolling out a big wireless switch and many small accesspoints under its control each of which has a contention ratio of under 1:10. At that contention ratio deployments for anything more than 10PCs is uneconomical.

This is all of course if we leave the security aside. But that is another story.

Re:Um, no.

[bleh-of-the-huns]

Yes, its called STP instead of UTP...

Re:Um, no.

[fbjon]

Not to mention: until someone figures out a way of turning radio from shared to switched medium for cheap, it ain't replacing anything.

Re:Um, no.

[Tiger4]

I agree that wired LAN is more secure than WiFi. But can't you do some pretty scary signal reconstruction by reading electromagnetic noise coming off your network cable? It's my understanding that this can be done from X yards away, through walls, whatever.
Where X is 1 - 3 meters. If you are running a Must Be Secure network in a single cubicle of a hostile cubicle farm, or up against the wall of an apartment, you might have trouble. The vast majority of people are inherently secure, at least against this particular threat.

Guess you could always wrap your cable in tin-foil.
Shielded Twisted Pair will deal with this for you. It has been on the market for the past 2 or 3 decades. Maybe more.

Re:Um, no.

[somersault]

Although once you have physical access to any ethernet ports, it's a lot easier to get onto an ethernet than it is to get onto secured wireless. Just plug in and away you go..

Re:Um, no.

[swillden]

Now, I don't see much reason to string ethernet through people's homes, at least not most of the time.

I do. Media servers. Although the theoretical data rate of 802.11n is high enough for several HD video streams, in practice you only get a third of the theoretical data rate reliably, making it barely adequate for 1-2 streams. Start actually moving those files around to store, say, on a laptop drive for watching later, and you'll really find wireless inadequate.

You'll be fine, and 802.11n is probably way faster than any internet connection you might have.

There are places in the world where that is not true today, where 100Mbps Internet connections are common. I expect we'll see that even in the US, as fiber-to-the-home initiatives are rolled out. There's one in my neck of the woods, called UTOPIA. Right now they're only providing 10Mbps (symmetric), but the plan is to ramp that up to 100Mbps in the future.

Re:Um, no.

[MorderVonAllem]

If someone has access to a physical port then your security isn't worth a damn anyway.

Re:Um, no.

[somersault]

If they really want in, they're going to get in. Most people's houses are going to be pretty easily accessible compared to trying to crack WPA wireless encryption.

Re:Um, no.

[NeilTheStupidHead]

In my apt building, someone just moved in and brought an 802.11n supported router with them. All of a sudden the 10 or so 802.11g routers in the building have all but stopped working. I used to be able to pick up my wireless connection anywhere in the building, but now I can't see it at all if I'm more than two rooms away and the connection constantly drops. I've talked to some of my neighbours and they have been having the same difficulties.

Re:Um, no.

[NeilTheStupidHead]

Actually, twisted pair does a pretty good job of shielding itself. It's far easier to detect and reconstruct the video output to a computer monitor (CRT) and can be done from quite a distance (tens of metres). We did this in school once as a demonstration, using perhaps $50 of components from a supplier, $50 of common household items from Walmart and an old B&W television that was in scavanged from a storage locker. We made a simple directional antenna using a steel collander and were able to read a message displayed on the instructor's monitor: "This wouldn't work on an LCD screen"

Re:Um, no.

[modecx]

Yeah, trouble is, it's a little more obvious when you trip over a foreign CAT 5 cable at 2:00 AM, just to find it leading over to your neighbor's house. It's also a lot easier to wire said cable up to a 110v plug...

Re:Um, no.

[JCSoRocks]

Rofl, is network security the issue at that point? Once they're in your house they'll just jack your nifty computer, NAS, media center, etc. The security of your network doesn't matter. They can just pick up your box, take it home, and run a simple brute force cracker against it.

Now you could say that that argument doesn't apply to businesses but I'd say that the computers that were stolen from Wells Fargo a few years back would beg to differ. At the time I was a Wells Fargo client and I received a nice letter about how my personal information may be among the thousands of records on the stolen machines... I obviously left Wells Fargo. What kind of bank lets someone walk out the door with a computer!?

Re:Um, no.

[tooslickvan]

This UTOPIA you speak of sounds ideal. Does it come with fluffy white bunnies?

Re:Um, no.

[sam0vi]

I was thinking about something else: Gigabit Ethernet

Re:Um, no.

[mulvane]

Not true. I could drive up in a surveillance van parked a couple doors away with extremely sensitive gear and actually tap into your electrical system and read what is going on with your system. Not to mention EM transmissions. Wired networks decrease this ability, and fiber all but eliminates it without some very sophisticated splicing tools. All in all, a wired network as far as security goes does improve it. The only way wireless will be secure if the wireless link itself is secured via encryption, and the data passed to the transmitting wireless point is already encrypted via vpn or some other link level encryption. That way even if the wireless link itself is broken, the data itself still has to be cracked. Another thing with wireless is that if someone is sniffing your traffic, you have no way to know. With cat5/6, you could if required do weekly/monthly checks on the wire to find if anything is suspicious. With fiber, you could know usually on the spot that it was spliced and someone had patched in for snooping purposes.

Re:Um, no.

[The -e**(i*pi)]

ever hear of managed switches
they not only can require registration before turning on a port, but allow only one specific mac per port and either notify you, record all the data, and or shut off any port that is doing anything funy with mac addresses.

they also have a VLAN capacility that makes virtual switches connecting any ports in your whole building so noone but those on those specific ports can listen in.

and hacking a CISCO switch is no easy task.

Re:Um, no.

[swillden]

It's much easier to eavesdrop on WiFi than it is on a wired network.

Not if the WiFi network is configured for reasonable security. Physical access is typically much easier to get than the AES keys.

Re:Um, no.

[nacturation]

Regardless how much thrust is applied/generated, a pig is simply the wrong shape. An unmodified pig can not fly or glide. But what if the pig were suspended via twine attached to the dorsal feathers of a flock of swallows?

Re:Um, no.

[cayenne8]

"If they really want in, they're going to get in. Most people's houses are going to be pretty easily accessible compared to trying to crack WPA wireless encryption."

I dunno about that....at least in the Southeastern part of the US. Once you get past the locked doors, you run the risk of dogs and the inhabitants with their guns drawn and ready to fire.

Somehow I think sitting out in a car a distance away trying to hack the wireless is a little safer. It does and the very least, make the 'head shot' a little harder.

:-)

Re:Um, no.

[Original Replica]

It looks like you need to repaint your apartment. Try this :http://www.safelivingtechnologies.ca/RF/Products_RF_Shielding_Paint_HSF54.htm Maybe if you just paint the wall that faces that neighbor, you will still get your signal down in the lobby.

Re:Um, no.

[InvalidError]

Security is one reason I prefer wired over wireless... and since most of my networked equipment stays put in one room, extra cables are non-issue.

The other reason is reliability: I can count on my 100BaseTX network delivering 7-9MB/s with very little chance of external influences causing my link to either slow down or die. With wireless, I am at the mercy of nearby interference sources including cordless phones, electrical appliances, various gadgets and other wireless networking equipment, any of which can cause the link to do a number of undesirable things from retraining to going down.

There are two reasons I got WiFi: 1) my previous router was dying and 2) I got a laptop. I only use WiFi with the laptop but whenever I do large transfers, I still hook it up to Ethernet since it is ~5X as fast and never goes down. 802.11g is good enough for internet access and moderate file copying with my two laptops so I most likely won't be bothering with 802.11n until my 802.11g router either dies or becomes a broadband bottleneck.

BTW, it is possible to eavesdrop on Ethernet without touching the cables by capturing EMI from the UTP cables - there was a proof of concept for this some years ago where they managed to reconstruct a B&W image from a VGA cable by placing the receiver antenna ~1m from the cable using commodity components. That's pretty far from monitoring from a van parked a few houses down the road but it certainly proves the feasibility.

Re:Um, no.

[mcmonkey]

It's much easier to eavesdrop on WiFi than it is on a wired network.

Not if the WiFi network is configured for reasonable security. Physical access is typically much easier to get than the AES keys.

It's not just about ease of access, it's also about detection. It's very easy to break a window to get into a building, but it is also very easy to detect that type on intrusion. It may be more difficult to crack a WiFi connection, but it also not as obvious when someone is sniffing your packets.

And in any case any security comes down to the weakest link. If that link is physical access, when it comes to the point that I'm in your house standing in front of your computer, how is your security aided by using 802.11n over ethernet? (Other than one less wire for me to unplig before taking your system?)

It seems to me, wireless networking has all the security issues of ethernet and then some.

Re:Um, no.

[mcmonkey]

]This UTOPIA you speak of sounds ideal. Does it come with fluffy white bunnies?

No, but it does come with OMFG ponies.

Re:Um, no.

[alienw]

I don't think you could do that with a network cable. CRT monitors emit tons of RF, at very low frequencies (which propagate much better through walls). They have practically no shielding. UTP cable behaves as a transmission line (so it doesn't radiate RF all over the place), and uses pretty small signal levels. With very good equipment and a low-EMI environment, you might be able to do it from 1 or 2 meters away, but not much farther.

Contrary to popular belief, Faraday cages do little to stop professional snoopers. Unless you put all your stuff into an RF test chamber with no openings whatsoever, the effectiveness of your shielding will be pretty low. Professional spies can probably overcome the 10-20dB attenuation that you would get, just by using better equipment.

In most cases, it is far easier to gain access by purely physical methods. They can open doors, replace or bribe security guards, install surveillance equipment, and so on. There is no real way for any given individual to prevent any of that, so the whole argument is pretty pointless.

Re:Um, no.

[sumdumass]

It is sort of ridiculous to say that anyone could just break in and steal the computer. They could do that for a wired and a wireless network. That isn't a fact that should be part of the discussion.

Now with a wireless connection, a person could hide their attempts to access the network. They could also do it from far enough away to not impose suspicion. This ability is an extra ability that makes wireless dangerous to some on certain networks. It might not be the same for mom and pop, but when they are doing their accounting and managing banking acounts/passwords and whatever, it might.

It wouldn't be too hard for me to create a proxy server or DNS server on a network and use DHCP in order to issue a new DNS number and intercept log on attempts to their banks, credit cards or health insurance. Then All I would have to do is access those sites from another hacked wireless connection and start taking money. It is more involved when doing this with a wired network.

In the end, it is like the saying that locks only keep honest people out. Well you would be amazed at how many honest people aren't exactly honest when they think no one is watching them. The fear of getting caught doing something bad is enough to stop them from doing it. With a wireless network, you are somewhat allowing these honest people be in a position that no one is watching. But more appropriately, you are somewhat allowing the dishonest people inside your building unmonitored to some extent. Wire a wired network, sure it can still be done, it is just that the possibility of being caught is so much more aware to the honest person.

Re:Um, no.

[terminal.dk]

All this Mac address crap is no important for a hacker, he just clones the MAC address of a LAN PC. He can then use another IP address to avoid detection by the PC. That is not anything funny, and something the switch will not react to. It is perfectly normal to use an extra IP address. Or he can plug LAN PC Hacker PC network.

We use VLANs, and many of them. Yes, you can only see traffic from your local segment. And hacking the Cisco switch is no more difficult than hacking everything else.

In big places, they are limited so a few adddresses can manage them. Just spoof this on the uplink port if you get access to the box. And if they use an upstream radius server, this is easy to fake as well. Cisco is only secure as long as the infrastructure is physically protected. Same as for a PC.

Re:Um, no.

[NeilTheStupidHead]

Thanks for the tip, but at $475 per five litres, it would be cheaper to build a faraday cage, or buy a new, more powerful router and start an inevitable arms race that will eventually lead to violations of CPC-2-0-03. (Hint, since I build and repair transceivers for a living, I'll probably win ^>^ )

Re:Um, no.

[sumdumass]

Now wait a minute. I can sit along the street and with some antennas, you wouldn't even be able to see me from your building. In that case wired is safer. When you look at all the pros and cons verses security, and weed out what would effect both types of network, Wired beats wireless hands down.

Saying that someone could break in and steal the computers doesn't mean throwing any other thought of securing the network away. It could mean that you might be hacked by someone who would never break into anything or steal a physical object. I mean why even lock the house if they could break in anyways?

Your neighbor might never think of breaking into your house and stealing something from you. But he might not think twice about hacking into a wireless network that he can pick up from the safety of his own home. A lot of times, to the criminal, it is how likely they would get caught and how easy it would be to get away with it. How likely is it that you or I would be caught in the privacy of our own home with a wireless signal? How hard would it be compared to breaking into a house and stealing the computers?

Re:Um, no.

[MojoStan]

Have you asked the new neighbor to update his/her firmware? I'm not sure if this will help, but the new 802.11n Draft 2.0 certification is supposed to prevent interference problems with legacy 802.11b/g gear. Some, but not all, "draft" 802.11n routers can be upgraded to Draft 2.0 with a firmware update.

From the Wi-Fi Alliance's Draft 2.0 FAQ (PDF file):

• I heard 802.11n can cause interference problems with other Wi-Fi networks. Is this true?

  In some configurations, 802.11n products can interfere with other Wi-Fi networks when they are trying to achieve the best performance. However, all products that are Wi-Fi CERTIFIED 802.11n draft 2.0 are required to implement a good neighbor protocol that helps ensure that interference is not a problem. Wi-Fi CERTIFIED 802.11n draft 2.0 products will operate in a manner designed to cause the least interference.

Specifically, Draft 2.0 access points are supposed to switch from 40 MHz to 20 MHz when it detects an 802.11b/g neighbor. However, according to a SmallNetBuilder blog entry, some early Draft 2.0 certified gear don't implement this protocol correctly.

Re:Um, no.

[JoelKatz]

Um, no.

Where's the "-1 Wrong" modifier?! Actually, this one is wrong for so many reasons I don't know where to start.

The simplest thing to point out if that if you use a one-time pad more than once (and you're going to send more than one packet in 10 seconds, I assure you), you lose the security properties of the one-time pad. So all your syncing (which is obviously going to be a huge pain in the butt) is wasted since you didn't get the thing that it was supposed to get you.

OTP are essentially useless in practice. 99% of practical systems that claim to use OTPs actually don't. Worse, OTPs actually *don't* provide many critically-needed security properties and they magnify some vulnerabilities. (It's easier for a MITM to flip a specific bit of a packet protected by an OTP than for a packet protected by DES.)

wait

[wwmedia]

didnt they say the same about 802.11g not too long ago?

and what do we have now? both systems coexisting with each other

same gonna happen again

No

[DJ Jones]

RJ45 jacks will never be obsolete for one reason... Security.

And I don't know what you're talking about, I still use gaslights.

Re:No

[DrSkwid]

> RJ45 jacks will never be obsolete for one reason... Security.

Just like BNC

Re:No

[Ajehals]

Hmm, so what you are saying is that a badly configured wired solution is more secure than a perfectly configured wireless solution. Well that is fairly obvious. But is a perfectly configured wired solution less secure than a perfectly configured wireless solution? No its not.

If I have decently configured switches / AP's and correctly configured network services (DHCP, Cryto, IDS etc..) then which is more secure? Well the wired solution is. Why? because I am not making my network available for anyone to try and break into, or simply to monitor and carry out analysis against. Not to mention that the wired network will be faster, easier to troubleshoot (generally), generally cheaper (especially for big implementations) and easily managed. Wired networks have their place, as do wireless networks, the only people pushing for wireless everywhere are those with a stake in selling or providing wireless networking, sadly the idea appeals to a fair number of people (its very "high tech"), and 99% of Wireless networks I have stumbled across (pun intended) where not even close to being secure.

Re:No

[AK Marc]

The wireless solution is. Physical access is typically very easy to obtain. Much easier than cracking AES or finding a way to get a copy of the WPA PSK or an account in the EAP server.

Physical access is useless in a properly configured wired network. Have you ever heard of 802.1x? It's still considered a part of the best that wireless has to offer for security. Do you know what 802.1x was created for? Wired networks. If you properly configure 802.1x on a wired network, you not only need physical access, but you need to authenticate to be able to reach anything except the authentication server. You will get no broadcasts forwarded to your port for sniffing. You will not be able to send anything out of the network. You don't need encryption if you don't forward a single packet to that port other than the deny messages from the authentication server. Of course, you can encrypt over ethernet as well, but it isn't necessary with a properly configured network, even without physical security to the ports.

Wireless took Ethernet's leftovers and did the best they could, and it is still less than Ethernet. Equally secure networks *always* leave the wired network more secure. Anyone that thinks otherwise doesn't know how to configure a wired network.

Shared medium.

[saintlupus]

I can't wait for wireless to take over everything. Collisions and shared bandwidth are awesome. I miss hubs so much.

--saint

Yes, of course

[blueZ3]

This is totally a replacement for wired connectivity, because in a building with three or four hundred computer users, there won't be any radio interference between wireless cards. I'm sure that there won't be any issues in high-density deployments. I mean, the four PCs in my house never, ever have any reduction in speed when they're all connected simultaneously.

What do they teach them in schools these days?

Re:Yes, of course

[winkydink]

What do they teach them in schools these days?

Had a look at Myspace or Facebook? Sigh.

Re:Yes, of course

[peacewon]

Hmmm... Makes you wonder about the effects of multiple sequences of ones and zeros traveling through my brain right now in radio frequency form. Also makes me think of the fried egg drug commercial.

Re:Yes, of course

[Tiger4]

This is totally a replacement for wired connectivity, because in a building with three or four hundred computer users, there won't be any radio interference between wireless cards.
And when someone decides to fire up the microwave oven for coffee break or popcorn, all those other users can take a break with them. "Teaming" at its finest. This is Win-Win for all concerned! Synergize forward to the Wireless Productivity Revolution!!

Re:Yes, of course

[Tiger4]

A single antenna can deal with 200+ computers.
It isn't an issue of numbers per antenna, it is view angles, polarization, signal propagation and blockages. 4 antennas per AP lets you set them at different locations (wired back to the switch) and at various inclination angles to be sure the signal covers the space you care about.

Doubt it.

[ynososiduts]

Unless buildings are made of less concrete and brick. My school has a wireless network, but it's spotty due to the big maze of concrete and brick buildings. You only get a connection when the room you're in has a wireless bridge, but every room has a RJ45 port. There really is no question of signal strength when talking about wired networks.

Yeah OK

[betterunixthanunix]

Wireless will never beat Ethernet, if for no other reason, simple reliability. I have seen odd things happen with radio waves, like have a very good signal in one spot, and almost no signal just a few feet away. Or getting the signal strength affect by where some random person is standing. Or signals not passing through walls (getting a cable through a wall requires no more than a drill). Or microwave ovens killing the signal.

The strangest was a friend who used a linksys router with the SSID "linksys" and WEP encryption, who lived next door to someone using the same SSID but no encryption. Oh yeah, the wireless network managers on various OS's had a field day with that one. Ethernet just doesn't have those problems, so it will always been needed when mobility is less important than reliability.

Reasons not to change.

[GodCandy]

First I will admit. I have an 802.11n setup at my house for my laptop and a desktop on the far side of the house. It works well for this.

The issues are as follows.
Security: There is little or none. All of your transactions are flying through the air and anyone with the proper equipment (which can be obtained at the local electronics store for very little money) can intercept those packets. Even if you bother to use encryption all that has to be done is some processing to "crack" the encryption. Without breaking into my house/office and tying into my physical copper network there is no way to intercept packets on a copper network.

Stability: I cant speak for 802.11n as of yet. My AP has never been rebooted and my clients stay conected. However my prior 802.11x products were somewhat less stable.

Speed: 802.11x is a bus topology much like a hub. True they are running a great deal of bandwidth now. For few users this is great however what happens when you have 20 users on the same access point sharing the same bandwidth.

I do however see uses in business for this. I don't think at this time it is the end all replacement for the simple switch and the complicated wiring closet yet.

And how about open drivers.

[ahfoo]

The security doesn't bug me at all compared to the issue of open drivers. If all the drivers for 802.11n products were as open as wired ethernet then it would be an almost maybe possibility but as we've seen with regular Wifi, there's no way in hell. Personally, I think pushing yet more closed and fucked up drivers is almost certainly one of the goals of the 802.11n standard.

It's a well known fact that UWB and other existing techniques can push wireless bandwitdth far past what 802.11n offers, but they're not "ready" for the consumer market. The game is to incrementally push the consumer market into a series of screwed up proprietary drivers to push out open standards and ensure that only "enthusiasts" use open source.

Re:And how about open drivers.

[nine-times]

The game is to incrementally push the consumer market into a series of screwed up proprietary drivers to push out open standards and ensure that only "enthusiasts" use open source.

It's possible that you're being paranoid on the drivers issue (sorry, don't mean to be insulting, but it's possible). However, I'm sure that the game is to incrementally push the consumer market to new devices. Many hardware companies do this-- they don't want to release a real solution all at once, but instead constantly release incremental improvements. The game is to get you to keep buying more hardware.

Ubiquity

[Alzheimers]

It will only be truly ubiquitous when it's a common check box feature on every PC sold, built-in to the motherboard and included in the final price.

As long as it's a peripherial, I don't care how cheap or easy to install, it'll never replace what's already there, ie. the Ethernet port. For more reference, see USB vs. Firewire.

2.4GHz Hell

[pcjunky]

I once got a call from a client who said her WiFi wasn't working in her study. When I got their I found she was using a bluetooth mouse, 2.4GHz cordless phone, Wireless video extension (also 2.4GHz), and cooking diner in her microwave (big 2.4GHz transmitter). This piece of spectrum will only take so much. She asked if changing to a 5.8GHz phone would help. I said probably not as most transmit from the base to the phone on 5.8Ghz and the phone transmits 2.4GHz back. (900MHz would be better). As we use more and more 2.4GHz wireless stuff the performance of WiFi will drop.

A 100MBPS wired network with a switch will outperform any wireless network for the foreseeable future.

Don't forget denial-of-service

[coyote-san]

Don't forget that there are multiple aspects to security. You don't want the sleazy competitor sniffing your network, but you don't want them blasting your network out of existence two days before the RFQ is due either. The bad actor could be hard to track down if they're using a highly directional antenna and an illegal amplifier.

Not even mom-and-pop businesses

[DogDude]

No, Ethernet isn't going anywhere, especially for "mom-and-pop" businesses. Why? If you have a small retail store with two cash registers, and your wireless connection acts up, you have -zero- income until it's fixed. That's pretty mission critical. On the other hand, if you're IBM, and some of your wireless goes flaky, IBM isn't going to shut down. A lot of people will be inconvenienced, but very few parts of a giant corporation are mission critical. If anything, I'd expect to see wireless at big businesses, where there's a ton of redundancy already (again, IBM is a prime example). Small businesses' computer systems are much more critical, so I can't imagine any successful small business using wireless anytime in the near future. I know that mine isn't!

As a physicist...

[drolli]

i have to make a comment. Let's assume we have amplifiers and signal generators avaiable which have at each time a certain "eqi-cost" line in the "power consumption" vs "noise level" plane in the bandwidth you are interested in. If you couple these by an really good directed radio link, you can get over a few meters up to a few dB if you are good. Hovever lets assume fo the second that having a parabolic antenna on you laptop is less handy than an ethernet cable. Thus, leaving aside obstacles, wou will definitely have less power at the receiver for the same power send. now here comes the problem. Less power means lower signal/noise ratio, which directly reduces your BW. So no matter how the wireless standard looks like, if you take it literally you can always use it on a network cable, and you will get a much higher rate and an ultimately directed transmission. Nowadays etherenet standard does not use the full bandwidth of the cables. WOuld one use the wireless transmission methods on a cable, one could get substentially more troughput.

O did i forget? eqi-cost can also be translated to "cheaper modules" at the same rate.

Shared bandwidth

[gravis777]

The problem that clients in our building seem to neglect is that, yeah, while we are running G, which is 56Mbps, that does not mean that it will be only half the speed of their 100 Mbps ethernet connection, its generally much slower. The problem is, on ethernet, you have a 100 Mbps connection straight to the switch, dedicated to you. Over the wireless, you are sharing that 54Mbps connection with 50 other people in your area, so you are not getting 54 Mbps, you are getting between 1-5 Mbps. This is why you ge an excellent signal, then almost cannot browse the internet. i think we finally got it through most of our users minds that the wireless was there as a convienince, not at a replacement for the ethernet, and most will now use their ethernet cable.

Re:Shared bandwidth

[Jeremy Erwin]

Over the wireless, you are sharing that 54Mbps connection with 50 other people in your area, so you are not getting 54 Mbps, you are getting between 1-5 Mbps. This is why you ge an excellent signal, then almost cannot browse the internet.

So, are people at your company so accustomed to browsing wit personal T3 lines, or what? Will they break down in tears if forced to work with 768 kb/s DSL?

Re:Shared bandwidth

[tehcrazybob]

I can't speak for his company, but at mine there's a great deal more to the network than internet access. All of our files are stored on network servers. Since we are all connected via gigabit ethernet, these shares are as responsive as local hard drives. Our work would continue undeterred if we were limited to browsing the internet on 768 kb/s DSL, but if we started connecting to our server with the DSL I think productivity might drop. In fact, you're right, I might even break down in tears.

For reference, gigabit ethernet offers theoretical 128 MB/s transfers, while local hard drives offer between 60 and 90 MB/s. Obviously the latency will be a bit higher on the networked drives, but you'll see no drop in sustained transfer rates. Compare that to a theoretical maximum of 37.5 MB/s for wireless N or 6.75 MB/s for wireless G, and bear in mind that those speeds will be shared with all clients rather than dedicated as with the ethernet connection.

3 words

[MSDos-486]

Power over Ethernet

Re:I want my ETHERNET!

[pla]

Wireless is far more secure than wired. To listen on your wired network all I have to do is get access to a cable. To listen on your WPA-secured wireless network I have to get access to a copy of your WPA key (assuming PSK for simplicity, but similar difficulties apply to the other modes).

One of those you can do from the parking lot (or with a good antenna, quite a good distance away). One you need physical access for.

Thanks, but I'll run encryption over my wires before I'll switch to trusting the same broadcast to everyone in the area.

tcp vs udp

[sgt scrub]

Wireless will displace wired in the same way that UDP displaced TCP.

Re:I want my ETHERNET!

[pla]

How do you get the WPA key from the parking lot? Please, do tell. You can become quite famous and probably make some good money if you can answer this question.

Just as soon as someone finds the answer to that, or more likely, finds a way to get around needing it (let's not insult each other and pretend it will never happen), they can have the fame, I don't want it.

I understand your point, but it doesn't change the fact that, however strong you claim cryptosystem-X, I can still assert with 100% accuracy that running it over wires instead of broadcast RF greatly improves that strength.

Re:UTP vs STP (sheilded twisted pair)

[afidel]

Only if you use an STP compatible switch (rare) otherwise you have just created a giant aerial. That was just one of the things I routinely found when I was a consultant, people figured the cables were more expensive and said shielded so they MUST be better.

Re:The thing I find funny. . .

[AJWM]

the word "Ether" inclines one to think of sending messages through a mysterious invisible medium which connects all things in space

No, you're thinking of "Aether" (as in "lumineferous Aether"), whose existence was shown unlikely by the Michelson-Morely and follow-on experiments.

Ethernet is talking about "ether", the class of compounds where e.g. two alkyl groups are linked with an oxygen atom in between (eg diethyl ether). The network tubes are filled with this stuff. You might think that the reason is ether's high volatility means signals can go faster, but the real reason is far more subtle than that.

Take a look at the diagram of molecular structures here. The one at the top is ether. Now, what does that remind you of? Right! RFC-1149, A Standard for the Transmission of IP Datagrams on Avian Carriers. (Not to be confused with Evian carriers -- filling those tubes with water doesn't work at all well.) Being so much smaller (many orders of magnitude) than, say, Columba livia , those little ether molecules can travel a lot faster, with a corresponding increase in bandwidth.