Email Delivery

Receive new posts as email.

Email address

Syndicate this site

RSS | Atom

Contact

About This Site
Contact Us
Privacy Policy

Search


November 2010
Sun Mon Tues Wed Thurs Fri Sat
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30        

Stories by Category

Basics :: Basics
Casting :: Casting Listen In Podcasts Videocasts
Culture :: Culture Hacking
Deals :: Deals
FAQ :: FAQ
Future :: Future
Hardware :: Hardware Adapters Appliances Chips Consumer Electronics Gaming Home Entertainment Music Photography Video Gadgets Mesh Monitoring and Testing PDAs Phones Smartphones
Industry :: Industry Conferences Financial Free Health Legal Research Vendor analysis
International :: International
Media :: Media Locally cached Streaming
Metro-Scale Networks :: Metro-Scale Networks Community Networking Municipal
Network Types :: Network Types Broadband Wireless Cellular 2.5G and 3G 4G Power Line Satellite
News :: News Mainstream Media
Politics :: Politics Regulation Sock Puppets
Schedules :: Schedules
Security :: Security 802.1X
Site Specific :: Site Specific Administrative Detail April Fool's Blogging Book review Cluelessness Guest Commentary History Humor Self-Promotion Unique Wee-Fi Who's Hot Today?
Software :: Software Open Source
Spectrum :: Spectrum 60 GHz
Standards :: Standards 802.11a 802.11ac 802.11ad 802.11e 802.11g 802.11n 802.20 Bluetooth MIMO UWB WiGig WiMAX ZigBee
Transportation and Lodging :: Transportation and Lodging Air Travel Aquatic Commuting Hotels Rails
Unclassified :: Unclassified
Vertical Markets :: Vertical Markets Academia Enterprise WLAN Switches Home Hot Spot Aggregators Hot Spot Advertising Road Warrior Roaming Libraries Location Medical Public Safety Residential Rural SOHO Small-Medium Sized Business Universities Utilities wISP
Voice :: Voice

Archives

November 2010 | October 2010 | September 2010 | August 2010 | July 2010 | June 2010 | May 2010 | April 2010 | March 2010 | February 2010 | January 2010 | December 2009 | November 2009 | October 2009 | September 2009 | August 2009 | July 2009 | June 2009 | May 2009 | April 2009 | March 2009 | February 2009 | January 2009 | December 2008 | November 2008 | October 2008 | September 2008 | August 2008 | July 2008 | June 2008 | May 2008 | April 2008 | March 2008 | February 2008 | January 2008 | December 2007 | November 2007 | October 2007 | September 2007 | August 2007 | July 2007 | June 2007 | May 2007 | April 2007 | March 2007 | February 2007 | January 2007 | December 2006 | November 2006 | October 2006 | September 2006 | August 2006 | July 2006 | June 2006 | May 2006 | April 2006 | March 2006 | February 2006 | January 2006 | December 2005 | November 2005 | October 2005 | September 2005 | August 2005 | July 2005 | June 2005 | May 2005 | April 2005 | March 2005 | February 2005 | January 2005 | December 2004 | November 2004 | October 2004 | September 2004 | August 2004 | July 2004 | June 2004 | May 2004 | April 2004 | March 2004 | February 2004 | January 2004 | December 2003 | November 2003 | October 2003 | September 2003 | August 2003 | July 2003 | June 2003 | May 2003 | April 2003 | March 2003 | February 2003 | January 2003 | December 2002 | November 2002 | October 2002 | September 2002 | August 2002 | July 2002 | June 2002 | May 2002 | April 2002 | March 2002 | February 2002 | January 2002 | December 2001 | November 2001 | October 2001 | September 2001 | August 2001 | July 2001 | June 2001 | May 2001 | April 2001 |

Recent Entries

In-Flight Wi-Fi and In-Flight Bombs
Can WPA Protect against Firesheep on Same Network?
Southwest Sets In-Flight Wi-Fi at $5
Eye-Fi Adds a View for Web Access
Firesheep Makes Sidejacking Easy
Wi-Fi Direct Certification Starts
Decaf on the Starbucks Digital Network
Google Did Snag Passwords
WiMax and LTE Not Technically 4G by ITU Standards
AT&T Wi-Fi Connections Keep High Growth with Free Service

Site Philosophy

This site operates as an independent editorial operation. Advertising, sponsorships, and other non-editorial materials represent the opinions and messages of their respective origins, and not of the site operator. Part of the FM Tech advertising network.

Copyright

Entire site and all contents except otherwise noted © Copyright 2001-2010 by Glenn Fleishman. Some images ©2006 Jupiterimages Corporation. All rights reserved. Please contact us for reprint rights. Linking is, of course, free and encouraged.

Powered by
Movable Type

« How Draft N Makes Nice with Neighbors; 5 GHz Averts Tragedy of the Commons | Main | Fon Promotes Fonbucks »

February 16, 2007

AirPort Extreme: Apple Breaks 90 Mbps

My review of the new AirPort Extreme Base Station is up at Macworld: This lengthy review, aided by several colleagues at the magazine, covers a lot of the basics for home users. I gave the unit 4 1/2 mice for how well it lived up both to its potential and how well it works. I was able to see consistently high speeds in testing, in excess of 90 Mbps in a single direction over 802.11n to Ethernet (flooding packets from N to Ethernet), and about 50 Mbps when flooding from N to N via the base station. My conclusion is that the device really needs gigabit Ethernet to achieve its full potential.

You'll note that the AirPort Extreme is what I was referring to in a post a few days ago in which I described how I developed new testing methodology for Wi-Fi gateways. The Extreme has a minor flaw that won't bite many people in its ability to pass traffic at full Ethernet speeds across its WAN port when network address translation (NAT) is engaged. Apple said they are looking into the problem, which is software based. A source unconnected with Apple provided convincing proof that the AirPort Extreme uses NetBSD as its embedded operating system, and that the network stack in that OS could be at fault. But it could be trivial to fix, too. (Update: Not to be obscure about NetBSD: the Acknowledgements.pdf file found on the CD-ROM that ships with the AirPort Extreme provides full copyright and acknowledgments credit for included software, as required by a host of GPL and other licenses. NetBSD is thoroughly acknowledged there; the DHCP software is credited to ISC.)

I'll be writing more soon about particular aspects of the base station, but for now, I'd like to direct you to the technical discussion about the Extreme's use of IPv6, the next-generation Internet routing protocol that's been "next generation" for something like eight or nine years now. IPv6 support is found throughout Mac OS X and is fully supported in the Extreme base station--so fully, Ars Technica's Iljitsch van Beijnum reports, that by default every Mac OS X computer that connects to a new Extreme gateway will be fully reachable through tunneled IPv6 from the rest of the Internet.

9 Comments

Glenn,

Very helpful write-up. I would like to know if you still think it worthwhile to upgrade if I have an early MacBookPro (15" 1.83) which, as far as I know, does not have N capability. It seems that my computer is always left out of all documentation or reviews as if it fell into a blackhole.

Would appreciate any feedback.

Thanks

[Editor's reply: Only Core 2 Duo Macs (except the 1.83 GHz iMac) have N built in. It's probably not worthwhile to get an N router with *no* N devices in hand, unless you think the range would help. 802.11n can improve the range of G devices, but not nearly as much as with N adapters. If you need printer and file sharing centrally, then it's a good choice with some futureproofing as you upgrade equipment--gf]

I just recently came across your comments & reviews, very informative. I have the new Apple Airport Extreme arriving next week. There definitely is a learning curve with these new routers.

I have Buffalo's first Draft 'n router, the WZR-G300N. What I am seeing is that the manufacturers need to take a different approach in there documentation and do a better job of explaining how to configure these new routers so as to obtain optimized speeds. There is a lot of confusion & frustration over on the Apple discussion boards.

Glenn, is there any chance that you can provide an in-depth review on Buffalo's Wireless-N Nfiniti� Dual Band Gigabit Router & Access Point, WZR-AG300NH. I would like to see what your thoughts are on the router like you did on the AE and maybe a comparison between the two routers.

Thanks in advance.

[Editor's note: It's a good idea, and thanks for it. I have been reluctant until the IEEE vote last month to invest time in current generation Draft N gear because of my concern that it would have a kind of dead end. Since that's turning out not to be the case, it's probably time to roll up my sleeves.--gf]

What is the "convincing proof" that the AirPort Extreme runs NetBSD?

[Editor's note: Didn't mean to be obscure. I've clarified in the main post.--gf]

Dear Glenn,

Thanks for your great review at MacWorld. I still have doubts about the impact of legacy protocols on the throughput speed between the AEBS and n-computers. In its "Designing Airport Extreme Networks" document, Apple says each component achieves its maximum speed, and that seems to be my experience. In monitoring my network I can see that the connection between my MacBook Pro and the AEBS is much faster than the connection between AEBS and my old Airport Express running Airtunes. If I unplug the Express, I don't see any change in speed between by MacBook Pro and the new Extreme. It's a little unclear from your review whether the throughput drop you mention affects all components equally:

"However, add the transmissions of just one legacy AirPort client—one using the older b or g standards—on the same network and throughput plummets. In a test with two laptops, one containing an 802.11b adapter, and the other an 802.11n adapter, both transmitting data at full speed, overall throughput dropped to a range of just a few Mbps to 30 Mbps. But in more likely scenarios, in which a b or g computer is sending data only intermittently, throughput between n hardware should remain high."

It would be wonderful if we could resolve for certain the circumstances under which legacy protocols slow down an n-network and under what circumstances they don't. Then it would be easier to determine what to unplug. Just a suggestion. Thanks for your good work.

[Editor's note: Apple is certainly being optimistic in their description. If a single 802.11b device is transmitting at full speed, it would grab all the available bandwidth, even that capable of 20 times the throughput. So there's a dance between multiple devices that results in substantially diminished overall throughput in high load situations.--gf]

Glenn, how do you think the Airport Extreme compares to the Netgear WNR854T, which looks physically similar to the Airport, but more importantly has gigabit ethernet built in?

[Editor's note: Haven't tested that one yet. I've been resistant until the January IEEE meeting to test 802.11n devices, as I thought that it might be wasted effort. Now, the question is, should I wait until expected March 2007 firmware upgrades that will drastically change backwards compatibility (legacy interaction) and probably fix piles of bugs. The companies releasing routers have been doing quite a lot of interoperability testing, but they have released very few firmware upgrades in the last few months. -gf]

To Jeff's question about the effect of legacy devices on an 802.11n network:

One way that I've found helpful to explain this to people is that every device will get an equal number of data frames sent on the network (assuming all devices are running the same application). The problem with this is that there are technological restrictions that cause 802.11n data to move slower if legacy stations are present.

Let me try to give an example. As Glenn stated, Airport Extreme can reach 300 Mbps as its top speed. (Remember, that is the speed that one frame of data reaches traveling from AP to station. Application throughput will be much lower.) The problem is, to get that speed it must use technology that is incompatible with legacy Wi-Fi devices.

To be compatible with legacy devices, Apple's 802.11n devices must use legacy technology that reduces their top speed to 270 Mbps. Right there, you already have some slowdown on the network.

In addition to the data frame speed reductions, 802.11n devices must "warn" legacy Wi-Fi devices in the area that they are about to transmit. As Glenn mentioned, a wireless channel is shared by the AP and all stations connected to that AP. If an 802.11n device were to transmit without warning, legacy Wi-Fi devices could try to transmit at the same time, resulting in neither data frame being successfully transmitted.

The good news is that 802.11n devices have a more efficient way of "warning" legacy devices that they are about to transmit than 802.11 g devices had. (802.11g devices see about 40% overhead when warning an 802.11b station that is connected to the same AP.) To my knowledge there has not been extensive testing on the overhead caused by having 802.11n devices "warn" legacy Wi-Fi devices before they transmit, but you can assume that you are going to see at least 10% overhead there.

We've already dropped our data frame speed (300 Mbps to 270 Mbps) and we've added overhead from having to "warn" legacy Wi-Fi devices before we send data frames (probably at least 10%). That doesn't yet address Jeff's central question: how much speed is lost when legacy devices actually **send** data frames.

A good way to understand the effect to these legacy devices sending data is to kind of think of you wireless channel like a traffic intersection that uses stop signs. When 802.11n devices transmit, they zoom through at 270 Mbps. They take very little time to get through the intersection, thereby allowing other stations to reach their "turn" quickly. When a legacy Wi-Fi device transmits at 11 Mbps (802.11b) or 54 Mbps (802.11g or 802.11a), that clogs up the wireless channel. Other devices have to wait more time to take their "turn". If you compare Jeff's scenario of an 802.11b adapter, that's almost 25 times as much waiting that an 802.11b device is causing compared to an 802.11n device at top speed.

The end result is somewhat like what Glenn described. When an 802.11b device is continuously transmitting (think FTP, not VoIP), all other stations spend so much time waiting for their "turn" that your overall network speed ends up dropping to a level you'd expect from 802.11b networks.

[NOTE: Applications that support WMM solve this problem to some degree. With WMM a station or AP gets an equal window of time to send frames rather than an equal number of frames. This way an 802.11b station and an 802.11n station will cause everyone to wait their "turn" for an equal amount of time. The difference is, the 802.11n station will send up to 25x the number of data frames.]

The bottom line in all of this is that you want to upgrade your network interfaces to a higher speed standard **whenever your applications require it**.

For example, I used to just use Wi-Fi for Internet surfing, email, instant messaging and Skype calls. For those purposes, 802.11b was just fine because those applications don't require much bandwidth (and because my Internet service was 6 Mbps, which is just as slow as 802.11b). Now I've started streaming ripped DVDs from a desktop in my office to a laptop that I can hook up to my TV in my bedroom or living room. I found that I needed 802.11g or 802.11a (I chose 802.11a) to get the extra speed to stream those videos. Sometime in the future I expect to get an AppleTV as well, and at that point I will use 802.11n so that I can stream wirelessly to both TVs at the same time if necessary.

I realize I've gone into too much detail here respective to Jeff's original post, but it is a complex topic. Hopefully at the least this does an adequate job of describing the 3 ways legacy devices slow down an 802.11n network.

I liked your book on Take Control of Your AirPort Network, Glenn.
Is there going to be a new one out that is updated for the new Airport Extreme Base Station?

[Editor's note: Yes -- by the end of March. Production and schedules and timing have pushed it back a bit. I'll announce the new edition on this site. -gf]

Don't want to sound daft, but how would a Gigabit interface speed the throughput of the N to N connection? If it's typical for WiFi to not perform at full speed in both directions, unlike ethernet, and if that's the performance you're seeing, what's to improve upon, and how would faster ethernet ports accomplish it?

[Editor's note: I was definitely speculating, but it's clear that there's an internal mechanism that runs at 100 Mbps Ethernet speeds that preventing unidirectional speeds of higher than 90-odd Mbps with 802.11n. The fact that in each direction, I could achieve faster-than-half speeds, shows that whatever this mechanism is, it's restricting the speed. It's possible that it does some clever internal routing that provides a kind of switching among wireless signals as they pass through the device--that's wired switching of wireless signals.--gf]

In response to John's question about the Netgear WNR854T:

I've tested and used this unit and it has two problems. First of all, the connection and throughput to Broadcom (Mac Pro) wireless hardware is poor. I received approximately 25 Mbps from an N enabled wireless connection to a gigabit ethernet connection. That was a high speed. It went downhill from there. The second problem is that the logging on router is very basic. It does have an SPI firewall, which I'd like to see on the Apple base station. You can also turn off the wireless radio and use it as a wired router. One of the big advantages of Apple's unit is the support for the 5 GHz band.

I had hoped to use the Netgear WNR854T to replace a Gig-E switch and a Linksys wired router (BEFSX41 router/firewall/VPN). I ended up keeping the Linksys unit online because of its security features and getting the Apple AirPort Extreme Base Station due to performance with Apple-supplied wireless hardware.