Debbie's tests, results, and explanation roughly match our own. With mixed b/g clients the maximum aggregate throughput with most systems will be in the lower teens.
>From a previous WIRELESS-LAN posting: == Mixed mode is real problem. On downstream tests we saw performance drop 55 to 65%, from 25 Mbps per second to 11, 22 Mbps to 7, etc. This is aggregate throughput with one AP, and three b client and three g clients. The compatibility mode that King is mentioning is the use of CTS-to-self and other similar mechanisms to make sure that b clients don't jump when they're not to, as well as basic rates being transmitted in CCKM so that b clients can understand. More information can be found here: http://www.cisco.com/en/US/products/hw/wireless/ps4570/products_white_paper0 9186a00801d61a3.shtml == The results of our Meru and Cisco 802.11b/g co-existence tests can be found here: http://www.networkcomputing.com/gallery/2006/1109/1109f3b.jhtml It shows that Meru's 802.11b performance is poorer, due to the fact that Meru allocates each client an equal amount of air time, no matter what their link rate may be. This means that clients associated with a higher rate will of course be able to move more traffic over the same time period than one associated at 5.5 or even 1 Mbps. This exacerbates performance issues for clients at the fringe coverage area, but I believe Meru allows you to control this feature. The results of a previous enterprise WLAN test can be found here: http://i.cmpnet.com/nc/1603/graphics/1603f2c.gif and a description of how we tested it here: http://www.networkcomputing.com/showitem.jhtml?articleID=59301907&pgno=9 Kind regards, Frank -----Original Message----- From: debbie fligor [mailto:[EMAIL PROTECTED] Sent: Tuesday, June 19, 2007 3:04 PM To: [email protected] Subject: Re: [WIRELESS-LAN] B user in a G cell On Jun 19, 2007, at 14:36, Jorge Bodden wrote: > James, > > I've tested this theory with a B client and a G client on the same > AP and the B client works at B speeds and the G client works at G > speeds. We tested on both Cisco Aironet and Cisco LWAPP and both > yielded the same results. Someone out there may have had different > results, but it worked fine for us. we did this test when we did our cisco vs meru testing in the fall of '05. this was the newly aquired (then) thin airespace APs for cisco. here's the info for just that from the writeup I did at the time: b/g interaction -- Meru said they had impressive b/g inter- operability. To test this we setup an iperf client on a "g" connected laptop, and attempted to send 40 Mbps (udp). For Meru, when the "g" client was alone on the AP, iperf recorded the BW as being typically around 30 Mbps, when the "b" client associated with the AP doing a ping, it dropped by about 3-4 Mbps to about 27 Mbps. When the "b" client ping flooded something, it dropped to about 20 Mbps, and went back up when the ping flood was stopped almost immediately. With Cisco the "g" user was getting 25-30 Mbps BW, and as soon as the "b" client associated it dropped to 15 Mbps, and ranged as far down as 6 Mbps when the "b" client was actively doing things. During this test we made sure all other "b" clients were off in the room to be sure that it was a single "b" station connection causing the problem. Once the "b" client was turned off it took one to two minutes for the "g" client's speed to get back to the mid-20 Mbps. now the reason it works for meru (my understanding anyhow) is they take turns for the b and g users, and like David said the g data can go fast. I don't think there's been a change to require this in the standard recently, or the vendors would make a much bigger deal about it. Also, while I wouldn't put too much money in the absolute BW numbers from iperf, I think that the relative changes should be reasonably correct. but then I drank too much meru kool-aid too. > > Jorge > > David Gillett wrote: >> If there's a B user in the cell, the *control* traffic needs to >> be at B rates. >> During time slices given to G clients, it's not necessary that the >> *data* traffic >> be understandable by the B client.... >> David Gillett >> >> >> --------------------------------------------------------------------- >> --- >> *From:* Jamie Savage [mailto:[EMAIL PROTECTED] >> *Sent:* Tuesday, June 19, 2007 12:18 PM >> *To:* [email protected] >> *Subject:* [WIRELESS-LAN] B user in a G cell >> >> >> I always understood that 802.11G provides connection rates of 54 >> meg. but realistically has usable throughput of ~24meg. Also, if >> a B radio associates to a G AP then the usable throughput >> drops to >> ~8 meg. I was advised today that, due to recent enhancements >> (within the last year?), a B user in a G cell no longer lowers >> the >> bandwidth <24meg. for all G users in the cell. It doesn't sound >> right to me.....can anyone comment on these 'enhancements' (if >> they do exist?) >> >> ..........thx.................J >> >> James Savage York >> University Senior Communications Tech. 108 >> Steacie Building >> [EMAIL PROTECTED] 4700 Keele Street >> ph: 416-736-2100 ext. 22605 Toronto, Ontario >> fax: 416-736-5701 M3J 1P3, CANADA >> ********** 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/. > > > > -------------------- > > This electronic message is intended to be for the use only of the > named recipient, and may contain information that is confidential > or privileged. If you are not the intended recipient, you are > hereby notified that any disclosure, copying, distribution or use > of the contents of this message is strictly prohibited. If you > have received this message in error or are not the named recipient, > please notify us immediately by contacting the sender at the > electronic mail address noted above, and delete and destroy all > copies of this message. Thank you. > > ********** > Participation and subscription information for this EDUCAUSE > Constituent Group discussion list can be found at http:// > www.educause.edu/groups/. > ----- -debbie Debbie Fligor, n9dn Network Engineer, CITES, Univ. of Il email: [EMAIL PROTECTED] <http://www.uiuc.edu/ph/www/fligor> "My turn." -River ********** 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/.
