 superdogI Need A DrinkPremium,MVM join:2001-07-13 Lebanon, PA | This can't be true? Over at the UBNT forum, they are claiming that a Bullet HPm can get over 80 concurrent users when setup as an AP. This has to be impossible!. »www.ubnt.com/forum/viewtopic.php?t=13617 -- »www.wavecrazy.net
|
|
 iansltx join:2007-02-19 Golden, CO kudos:2 | Polling?
Seriously, I need to show this to my ISP. Maybe that'll get them to deploy new infrastructure  |
|
|
|
 | reply to superdog I really like Ub... but...
If you read the linked thread this guy, "Mike" is claiming 100M TCP. That is absolute bull sheite. It speaks volumes about the guy's qualifications. Why? Well because you can't get there from 802.11g/b/n. You can not do 100M on the wire with any non-proprietary RF protocol of which I am aware.
That said, it may be possible, but I stand by, "That is not possible on b/g/n protocol networks." Also, the man claims 80 subs. Sure, you might have 100 connected. Talk to me about throughput per user on something that actually has 80 subs connected.
I call, bull sheite on this one. |
|
 | Mike is a guy that works at ubnt. He is the head of the testing department if i remember right.
I believe the 100m is achived with compression. |
|
 | reply to superdog There are too many variables in the real world to make that statement as far as I'm concerned.
But I think he is also talking about 40Mhz channels (read SUCKS) and over subscribing 2:1 in his next statement. In general you can plug as many people as you want almost on an AP. It's the overall air time/available speed that matters.
I think I can have 200 or so (software limited) on my Star AP's. I can't tell you how fast of service I could provide doing so though. 
They are certainly doing some interesting things there though. |
|
 | reply to AnonDOG said by AnonDOG :I really like Ub... but... If you read the linked thread this guy, "Mike" is claiming 100M TCP. That is absolute bull sheite. It speaks volumes about the guy's qualifications. Why? Well because you can't get there from 802.11g/b/n. You can not do 100M on the wire with any non-proprietary RF protocol of which I am aware. That said, it may be possible, but I stand by, "That is not possible on b/g/n protocol networks." Also, the man claims 80 subs. Sure, you might have 100 connected. Talk to me about throughput per user on something that actually has 80 subs connected. I call, bull sheite on this one. I'm finding this really hard to believe as well. |
|
 | reply to superdog There is another guy - WHT on the ubnt forums. He is a bit of an expert over there and has been testing the new bullets. I would take his advice as he is critical where he sees fit so he would be happy to give you some real-world results.
I cant wait untill they confirm the polling - that will solve a couple of headaches for me. |
|
 kewlkeedGrouchPremium join:2005-02-05 Knowlton, QC kudos:1 | reply to superdog Bingo with everything said above.
I think there's some serious horseshit going on with Mike, and either he doesn't know what he's talking about or is dodging the bullet (No pun) and not wanting to answer critical questions that may make them look bad.
IF it is polling that they are using, then theoretically you COULD get that many per AP, and the CPU of the AP would handle the load, but again that's IF it's polling.
Since nobody at Ubnt seems to want to tell us what this Max really is, I'm not even going to toy with these (I have a crate sitting next to me). I'm not going to bench test something I don't know what it does, and worse to bench test it to find out what it is, when the godamn manufacturer should be the one to tell me/us what it is. -- Justin - DSLR resident grouch and Mr Negativity TSI Fanboy - "Dontchya wish your 'net was hot like mine! Ohhh Dontchya!" Have a nice day! |
|
 GNca GeorgeGorillaNET Wireless BroadbandPremium,VIP join:2008-07-12 Minden, ON | reply to superdog I wonder just how many clients were hooked to an AP during the beta test phase? Not a lab setting, but a real field test with a normal noisy environment etc?
Having used MikroTik for seven years, I suspect we may see some further refinement of AirOSV before we want to use it for big PtMP deployments. That would only be normal.
Once we finally find out what the hell Airmax is and how effective its going to be, this could be pretty interesting.
We probably wouldn't use it in 2.4 though, too much noise. The other problem is how to feed it. A 100Mb-capable AP (or worse, a cluster of them) needs a lot of feeding... -- Don't steal, Bell and the CRTC don't like competition!!! |
|
 | reply to superdog All the test I have seen,
N performs rather well with UDP, TCP not so good.
Mikrotik N using dual antennas and Mimo on a 40 Mhz channel yields about 180+ Mbps UDP, and about 60 Mbps TCP.
I would expect half of this on a Ubiquity M Bullet, they only use 1 channel.. no mimo advantage |
|
 | reply to superdog I have two 802.11b Mikrotik APs with over 60 users on each, customer plans go from 500 Kbps to 1.5 Mbps. We are just now getting complaints on the APs and have are trying to move customers over to our less used APs on that site. 80 users on an 802.11G/N I believe is possible but the experience might not be the greatest. -- CCNA, Comtrain Certified Tower Climber |
|
 GNca GeorgeGorillaNET Wireless BroadbandPremium,VIP join:2008-07-12 Minden, ON | reply to superdog There is a little more info about Airmax on a Czech web site. Fascinating reading and if it all works as intended it really is world changing. I made a pdf in case it suddenly disappears!
»i4wifi.blog.cz/0908/150-mbps-spo···ubiquity
George -- Don't steal, Bell and the CRTC don't like competition!!! |
|
 1 edit | reply to superdog Mike is pretty good at what he does guys. He's always been straight with me. I'm providing good service to my subs 50 to an AP with Mikrotik on 10Mhz channels. I don't think what he is claiming is that out of order.
Dobry! Hooray for learning a modicum of Czech! I can make sense out of this without Google translator. Hopefully this product does not turn out to be hovno. |
|
 | reply to superdog I'm not saying impossible by any means. But it's just like anything else. You will have to pay attention, lock rates, have ABSOLUTE minimum signal levels for installs. etc. etc. to get that many on an AP and provide high speeds. Some on this forum are doing 2-5Mbps client speeds already, but limited to actual users per AP from air speed + over subscription.
The fact we are talking about 100Mbps over the air is one of the biggest reasons it's even possible. That and the "N" differences over "B/G". It's pretty exciting really. BUT, it's still new. And I won't be jumping ship quite this moment to do a full on town to try it out.
LOL, I'm just starting to get my new towns up with half channel widths and G mode Which is awesome just by itself compared to "B-only". If you ditch "B" things get fun anyway it seems.  |
|
 kewlkeedGrouchPremium join:2005-02-05 Knowlton, QC kudos:1 | reply to superdog Wow that just blew the skirt up on Ubnt hehehe!
At least now I have an idea where it's going. Which as I thought I wouldn't be far off in guessing within a year all their 802.11a/b/g stuff is EOL.
Looks like some new fun equipment coming up. Also looks like they're basically going to start directly competing with Canopy etc with the proprietary stuff. Could be interesting 
Thanks Ubnt for the NON heads up, and thanks for someone leaking it anyway 
*Looks at Ubnt* So this is what all the hype-max shit is all about? Hoo-rah! How fucking hard was that to tell people. -- Justin - DSLR resident grouch and Mr Negativity TSI Fanboy - "Dontchya wish your 'net was hot like mine! Ohhh Dontchya!" Have a nice day! |
|
 | reply to superdog I appreciate the personal attacks guys. I do nothing other then try and help each individual user who uses our products to the fullest extent within reason. I dont post bullshit of fluff your skirts, just what I see in testing. I do not get anythign from the top telling my to skew my results. I have quite and extensive networking background with many, many certs plus years working as a network admin with both wired and wireless networks. I may only be 28 but I have been in the game for over 10 years.
100Mbps TCP/IP speeds HALF duplex are attainable. You can wiki the dfference betwee HAFL duplex and FULL duplex links if you like to expand your knowledge a bit.
802.11abgn is at HAFL DUPLEX operation.
10/100 is FULL duplex (so you can have ~90MbpsTCP/IP going) in EACH directions. So in theory the maximum threshold for HALF duplex oepration is ~180 or so MBPS TCP IP. You can test it yourself. Run chariot over a wire between two lapopts. Gaurantee with traffic passing in both directions you will get ~180Mbps TCP/IP aggregate throughput.
When maxing my links while doing outdoor testing this is with traffic passing in EACH direction as well. Not uni-directional traffic.
Also the people who are experiencing slow throughput are using incorrect settings, hence my guide to maximizing 11n performance outdoors. Yes this is a new technology and there probably will be bugs that I have not thought of to test for, but I do what I can, listen to my user's at ALL times, and move the firmware forward to a point where it needs to be.
Thanks,
Mike Ubiquiti Networks |
|
 GNca GeorgeGorillaNET Wireless BroadbandPremium,VIP join:2008-07-12 Minden, ON | There are some very experienced people on here Mike, as well as some not so experienced people trying to learn.
I started with Arcnet when you would have been about 5 for example and I count myself as a mere tyro compared to some on this forum.
I really appreciate the help you and Robert have provided, that problem in early 2007 was a nightmare and sorting it out helped us a lot. Since then we've deployed a lot of your gear, but mostly in small business and residential settings where I don't have much day to day involvement. Hopefully we will have more UBNT gear in our core corporate and MUSH business with the 11N product range.
UBNT hasn't done themselves any favors with this latest 'hide the Airmax' saga, but I don't expect any lasting damage. The product is far too interesting to bear grudges.
George -- Don't steal, Bell and the CRTC don't like competition!!! |
|
 kewlkeedGrouchPremium join:2005-02-05 Knowlton, QC kudos:1 | reply to superdog Just to clarify,
My skirt comment was meant more about blowing the lid off the "secret" AirMax, not as a personal attack. (I'm not sure if it was taken that way, just want to be sure)
I also want to be clear I've never doubted the ability per say, nor to do I feel a serious lack of trust towards the numbers. I'm sure the units will deliver what they are supposed to, might take some ironing in firmware, but they'll get there.
I've never had a beef with the technical side of Ubnt (I wouldn't be using them if I did), or with Mike (You do the best you can with what info you're allowed). My biggest and only beef is with whatever is screwed up on the admin/paperpushing side of things. Something or someone somewhere up there doesn't have their shit in line. -- Justin - DSLR resident grouch and Mr Negativity TSI Fanboy - "Dontchya wish your 'net was hot like mine! Ohhh Dontchya!" Have a nice day! |
|
 | reply to superdog
Ummm...OK, time to come out of the closet.
I am the infamous WHT.
I'm the cowboy on the left and that's Mike Ford next to me.
Yes, I have been running the Bullet 5M since April. I was one of two WISPs selected by Mike Ford at Ubiquiti to be beta testers under a non-disclosure-agreement.
The Bullet M will indeed support in excess of 100 user per AP. |
|
 | reply to superdog Look guys, as Mike has said his hands are tied by marketing dudes. He couldn't explain things he wasn't allowed to explain. That is not his fault. We don't write his paycheck. He answers to his boss not the lot of us.
He goes a pretty good way beyond what other company support guys do and has done a very good job answering our questions for $0 on a forum that does not belong to his company. Quit being pricks.
The acidity of this forum his growing more and more every day. It's becoming a turn off. |
|