Network Computing did some specific testing on bad-neighbor effects for APs collocated on the same channel here: http://www.networkcomputing.com/channels/wireless/showArticle.jhtml?articleI D=194400865
The graphic that shows the speed differences can be found here: http://www.networkcomputing.com/gallery/2006/1109/1109f3d.jhtml You can see how the Cisco AP suffered in comparison to the Meru AP. If one presumes that in the Cisco vs. Cisco that each Cisco AP took approximately half the air time, and in the Cisco vs. Meru test Meru claims they also took half the air time, why didn't Cisco do better in the dual-vendor test? My guess is that the half of air time that Meru leaves for the Cisco AP isn't grouped or timed well enough for it to get a higher throughput. I should add that NWC tested in full saturation mode, and so in light traffic conditions I wouldn't expect to see any measurable bad-neighbor effects. We also had some post-article vendor statements, too: http://www.networkcomputing.com/showArticle.jhtml?articleID=195900045 Meru's Wi-Fi Alliance (WFA) re-certification was nothing but a verification that their product interops with others. It does *not* verify strict compliancy with the IEEE standards, as much as their press release and marketing material would suggest. I've had extensive discussion with the WFA in regards to what they actually certify, and it's just interoperability and compatibility, despite their own whitepaper claims. As for compliance, the WFA tests just a limited subset of the full 802.11 standard (and addendums). What those exact elements are is something that the WFA will not disclose to non-members. In other words, just because the WFA certifies it doesn't mean that Meru isn't doing something funky that can lower throughput for APs collocated on the same channel. Kind regards, Frank -----Original Message----- From: Ruiz, Mike [mailto:[EMAIL PROTECTED] Sent: Monday, April 09, 2007 8:58 AM To: [email protected] Subject: Re: [WIRELESS-LAN] microcell vs virtual cell Id like to share two pieces of info on this. We have been running meru as a neighbor to several other smaller wireless installs and have never seen any issues that were unexpected. This is both in an overlapping and a non-overlapping channel scenario. Secondly, a short time ago we hosted an independent lab who tested for the bad neighbor issue. They were unable to find any problems. I would expect the wifi recertification should also speak worlds on this alleged issue. - Michael Ruiz - Sent using Exchange Mobile Active Sync -----Original Message----- From: "Lee Badman" <[EMAIL PROTECTED]> To: "[email protected]" <[email protected]> Sent: 4/9/2007 9:05 AM Subject: Re: [WIRELESS-LAN] microcell vs virtual cell I've heard a growing number of anectdotal instances where the virtual cell model causes problems for neighboring WLAN systems by trying to control their timing parameters and such- though can't say that I have talked to anyone directly that has experienced this supposed "bad radio neighbor" effect. Has anyone who actually uses the virtual cell hardware had reports from nearby systems of this negative effect, or is this a bit of a competitors' urban legend? Regards- Lee >>> [EMAIL PROTECTED] 4/7/2007 12:15 PM >>> We too are Meru users, since December 2005. Michael has done a very good job articulating details of the Virtual Cell. I would be pleased to provide information if needed. Additionally I would be pleased to talk offline about some interesting technology we are alpha-testing from Meru. For what its worth, I wouldn't recommend doing Wi-Fi any other way. In the interim, I recall some independent layer 1 testing and operational testing done out of the UK a while ago. I'm trying to track down that information. Mike - Michael G. Ruiz, ESSE ACP A+ Network and Systems Engineer Hobart and William Smith Colleges Information Technology Services P.315-781-3711 F.315-781-3409 Team Leader: Derek Lustig ([EMAIL PROTECTED]) Did you know that HWS Students, Faculty, Staff, Alums, etc can purchase computers, accessories, electronics and software at a discount through our partner CDW-G? http://www.cdwg.com/hws/ - ________________________________ From: Michael Griego [mailto:[EMAIL PROTECTED] Sent: Fri 4/6/2007 6:49 PM To: [email protected] Subject: Re: [WIRELESS-LAN] microcell vs virtual cell Where virtual cell deployments really shine is in a couple of ways: 1. By timing the transmissions of both the APs and the clients, they cut *way* down on the number of collisions and retransmits. This alone is what causes the throughput of a normal AP to completely tank after 20-30 users. So, by cutting down on the amount of waisted air created by the random backoffs and the collisions themselves, you gain quite a bit of usable throughput and the ability to reliably support more than 20 users (since the available spectrum can be equally divided without the clients fighting like a bunch of siblings). 2. By moving to an almost TDMA approach, 802.11g clients get better performance when 802.11b clients are sharing the cell than they would with traditional APs (at least this is true for Meru). This is because the AP will give each client the same amount of air*time* instead of the same number of frames, allowing the 802.11g client to transmit more data before again having to wait on another client. 3. Most people don't realize (or it just doesn't dawn on them) that you *can* run all 3 channels in a virtual cell deployment. You do have to install more APs to support this configuration, but, by doing this, you get 3 virtual cells spanning your campus and all of the available bandwidth that goes along with it (which, for the reasons listed above, is more than you would get using a traditional 3 channel deployment, making your actual aggregate available throughput much closer to the 162Mbps theoretical max for 2.4GHz usage). One of the other nice benefits of virtual cell deployments is the lack of client-initiated roaming. This is especially useful for cutting down roam times when the WLAN is 802.1x authenticated (and it doesn't require PMK). Since, even though the client has moved his association to a new physical AP, he's still talking on the same channel and to the same BSSID, he has no clue that he has roamed and his session state has been seamlessly moved by the controller. I'd be happy to discuss (offline) our Meru system with anyone who'd like to ask questions. --Mike On Apr 6, 2007, at 3:30 PM, Ringgold, Clint wrote: > I am interested in the findings as well. My concern is the actual > throughput. It would seem to me that a virtual 3 ap setup would be > 54MB > while in a microcell it would be 162MB....Potential. > > I hope I'm wrong and or can get clarification. > > > > -----Original Message----- > From: Scholz, Greg [mailto:[EMAIL PROTECTED] > Sent: Friday, April 06, 2007 3:59 PM > To: [email protected] > Subject: Re: [WIRELESS-LAN] microcell vs virtual cell > > I am also interested in anything you find. > > > -----Original Message----- > From: Steve Fletty [mailto:[EMAIL PROTECTED] > Sent: Friday, April 06, 2007 3:33 PM > To: [email protected] > Subject: [WIRELESS-LAN] microcell vs virtual cell > > Is there any scholarly or technical data/analyis of the single-channel > virtual cell architecture vs the traditional micro-cell WIFI > achitecture? > > I don't want to hear from vendors. I don't want bake-off results or > vendor white papers. I'd like to know if there's any hard science > comparing the two contrasting schemes. > > -- > Steve Fletty > Network Design Engineer > University of Minnesota > > ********** > Participation and subscription information for this EDUCAUSE > Constituent > Group discussion list can be found at http://www.educause.edu/groups/. > > ********** > Participation and subscription information for this EDUCAUSE > Constituent > Group discussion list can be found at http://www.educause.edu/groups/. > > ********** > Participation and subscription information for this EDUCAUSE > Constituent Group discussion list can be found at http:// <http:///> > www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/.
