Key Questions

[corky]

Hi, interesting looking product. I'd need to know a bit more to consider deploying though:

1) Does this technology work / work well with point to multi point? I notice on your web page is says optimized for backhaul.

1a) If so, could I use an SR71 in a Mikrotik for the AP, and bullets for CPEs?

1b) I notice that some of the SR71 series are listed as g/a/n devices. Can a single AP serve both the older generation and the newer generation clients at the same time on the same sector?

2) Can we still do 5/10/20/40 mhz channels, or is it only wide channels?

3) How does range compare to plain 802.11a/g?

[WHT]

1) Does this technology work / work well with point to multi point?
A - Works for PtP backhauls and PtM AP applciations.

1a) If so, could I use an SR71 in a Mikrotik for the AP, and bullets for CPEs?
A - You can use anything you want behind the backhaul.

1b) I notice that some of the SR71 series are listed as g/a/n devices. Can a single AP serve both the older generation and the newer generation clients at the same time on the same sector?
A - No, if you use traditional 802.11a/g non-N clients, it would force the Bullet M series to slow down. You can't mix and match, its "M" all the way.

2) Can we still do 5/10/20/40 mhz channels, or is it only wide channels?
A- It support5, 10, 20, and 40 MHz wide channels.

3) How does range compare to plain 802.11a/g?
A - There is NO comparison.

The Bullet 5M with a 27 dBi grid can reliably support a 65Mbps TX & 65 Mbps RX connection with 100% CCQ at 22 miles. Even at 105º summer afternoons, 62º spring nights, 2 miles of solid rain thunderstorm there is no change.

This was unheard of a mere year ago.

[leo_cakep]

wow nice... can't wait to try bullet5M....

[drwho17]

You can't mix and match? I just got around to replacing all my tranzeo's with Bullet2HP's, now I'm looking at replacing the Bullet2HP's with M's for the backward compatibility.

I'm not as much concerned about speed, as reliability and increased scalability with the improved CPU/mem. Do you just mean it will need to run in a protected mode for B/G clients, like G does for B, if so that's not a big issue, for me. I'm interested in putting more then 30 users on a AP.

[SatireWolf]

1) Does this technology work / work well with point to multi point?
A - Works for PtP backhauls and PtM AP applciations.

1a) If so, could I use an SR71 in a Mikrotik for the AP, and bullets for CPEs?
A - You can use anything you want behind the backhaul.

1b) I notice that some of the SR71 series are listed as g/a/n devices. Can a single AP serve both the older generation and the newer generation clients at the same time on the same sector?
A - No, if you use traditional 802.11a/g non-N clients, it would force the Bullet M series to slow down. You can't mix and match, its "M" all the way.

2) Can we still do 5/10/20/40 mhz channels, or is it only wide channels?
A- It support5, 10, 20, and 40 MHz wide channels.

3) How does range compare to plain 802.11a/g?
A - There is NO comparison.

The Bullet 5M with a 27 dBi grid can reliably support a 65Mbps TX & 65 Mbps RX connection with 100% CCQ at 22 miles. Even at 105º summer afternoons, 62º spring nights, 2 miles of solid rain thunderstorm there is no change.

This was unheard of a mere year ago.

I'm going to replace my primary backhaul with these and see if it cleans up my nano reset'itis.

Then maybe I can get them rma'd...

[WHT]

You can't mix and match? You can mix and match, but you loose the "M" features. If want to take advantage of the features, then your AP has to be "M" and all you stations also.

You can run an "M" backhaul and standard 802.11a/b/g WISP, and you'll have all the "M" advantages with the backhaul.

[drwho17]

You can't mix and match? You can mix and match, but you loose the "M" features. If want to take advantage of the features, then your AP has to be "M" and all you stations also.

You can run an "M" backhaul and standard 802.11a/b/g WISP, and you'll have all the "M" advantages with the backhaul.
What features would I use? Can I still run 10mhz wide channels? Can I support 802.11n/802.11g clients simultaneously? Will the units handle more subscribers per AP because of the improved CPU?

Don't get me wrong Bullet5-M's will be a dream for backhaul, but I'm hoping there is a path going forward to improve the areas where we've deployed Bullet2HP's for AP's. They have worked great and reliable thus far (some we have up for 5 months), no failure's on the tower's (we've had some fail out of the box). My biggest fear is scalability, which I'm hoping the M series will improve upon.

[WHT]

What features would I use? Can I still run 10mhz wide channels? Can I support 802.11n/802.11g clients simultaneously? Will the units handle more subscribers per AP because of the improved CPU?
You can use 5, 10, 20, and 40 MHz channel width.

I don't know details if an "M" AP will support a non-"M" radio, but I do know that I couldn't connect a Bullet M5 to a Bullet 5. However, it might be a simple thing that I had to change, I simply didn't have time to pursue it at the moment. I'll get around to that next week when I install another twenty Bullet "M" station system.

There are additional features that can't be released for advertising yet, and I'm not at liberty to discuss them at the moment.

[corky]

Thanks WHT for the very helpful answers! I have a few follow on questions, hope you don't mind.

1) Could you point me to a link that really gets into the exact details of the N protocol? I could not find one.

My main concern has to do with a limitation of 802.11a/g. Specifically, if a client on an AP is downloading at the APs full capacity, all other clients upload speeds drop to very low speeds. They jump around in the 0 - 200K range. Even worse, latency is severely impacted, which impacts VOIP.

I recently learned to resolve this with Mikrotik's newest version of nstream, but of course this makes CPEs much more expensive.

Soo...

2) Can nstreme work with the sr71 series?

3) Or better, does N resolve this problem natively?

[WHT]

My main concern has to do with a limitation of 802.11a/g. Specifically, if a client on an AP is downloading at the APs full capacity, all other clients upload speeds drop to very low speeds. That's a general networking limitation, you're have the same thing happen with a wired network.

[WHT]

2) Can nstreme work with the sr71 series? nstreme is a Mikrotik propriety format that can't be ported to UBNT devices.

[Shockware1]

Sounds like you need some form of bandwidth management to get control of your network. QoS may help some, but probably not enough. I suggest both get put in play.

You can't have one guy stealing the show from everyone else after all.

[UBNT-Mike.Ford]

Hi, interesting looking product. I'd need to know a bit more to consider deploying though:

1) Does this technology work / work well with point to multi point? I notice on your web page is says optimized for backhaul.

1a) If so, could I use an SR71 in a Mikrotik for the AP, and bullets for CPEs?

1b) I notice that some of the SR71 series are listed as g/a/n devices. Can a single AP serve both the older generation and the newer generation clients at the same time on the same sector?

2) Can we still do 5/10/20/40 mhz channels, or is it only wide channels?

3) How does range compare to plain 802.11a/g?

WHT Answered these correctly but just want to add:

1b) I notice that some of the SR71 series are listed as g/a/n devices. Can a single AP serve both the older generation and the newer generation clients at the same time on the same sector?

This is not recommended at all, as if you connect a legacy device you will force the unit to slow down to legacy speeds, negating the benefits of 11n.

Thanks,

Mike

[drwho17]

What about the scalability with the New CPU board? Is it reasonable to assume this will handle more users at G, then a 180mhz unit would?

[corky]

>> That's a general networking limitation, you're have the same thing happen with a wired network.

Hunh? This is not a problem with DSL, Cable or Nstreme wireless networks. On a typical wired LAN I've never seen any such behavior either. On a lan, all stations are peers, there is not an AP. Specifically, upload literally goes to 0 at times lasting a second or two on 802.11a/g networks when downloads are running full speed. If a second client wants to download, they the bw is shared reasonably, but uploads are very very bad. This seems to be a problem of outdoor wireless exclusively.

>> nstreme is a Mikrotik propriety format that can't be ported to UBNT
devices.

I have many Ubiquity SR cards in Mikrotik boards running nstreme. Ubiquity makes great hardware, and the Nstreme protocol works well with your SR cards to further improve performance. You make money either way :-) I just can't use bullets or nanostations with Nstreme, I have to use SR cards. Ubiquity SR cards and Nstreme make a good partnership. My question is if that type of partnership will still work with the SR71s.

[UBNT-Mike.Ford]

What about the scalability with the New CPU board? Is it reasonable to assume this will handle more users at G, then a 180mhz unit would?

Hello,

The untits will handle significantly higher loads then the previous Bullets as the process speeds are up from 180Mhz to 400Mhz on the new products.

Thanks,

[drwho17]

What about the scalability with the New CPU board? Is it reasonable to assume this will handle more users at G, then a 180mhz unit would?

Hello,

The untits will handle significantly higher loads then the previous Bullets as the process speeds are up from 180Mhz to 400Mhz on the new products.

Thanks,
Excellent, any testing been done on this? Currently my policy is to keep it below 30, w/about 20 at 768/128 and about 10 at 2/384. If I could double this per sector that would be a great value.

[UBNT-Mike.Ford]

What about the scalability with the New CPU board? Is it reasonable to assume this will handle more users at G, then a 180mhz unit would?

Hello,

The untits will handle significantly higher loads then the previous Bullets as the process speeds are up from 180Mhz to 400Mhz on the new products.

Thanks,
Excellent, any testing been done on this? Currently my policy is to keep it below 30, w/about 20 at 768/128 and about 10 at 2/384. If I could double this per sector that would be a great value.

Hey Drwho,

More info will be made available on this at a later date, but I can tell you that the client load is significantly higher then you may think.

Thanks,

Mike

[Mike_27]

do these devices support DFS and TPC in the FCC domain?

Mike

[raytaylor]

>> That's a general networking limitation, you're have the same thing happen with a wired network.

Hunh? This is not a problem with DSL, Cable or Nstreme wireless networks. On a typical wired LAN I've never seen any such behavior either. On a lan, all stations are peers, there is not an AP. Specifically, upload literally goes to 0 at times lasting a second or two on 802.11a/g networks when downloads are running full speed. If a second client wants to download, they the bw is shared reasonably, but uploads are very very bad. This seems to be a problem of outdoor wireless exclusively.


Could this have something to do with the rts/cts and packet size threshold settings? Could be just getting hidden node or packet collisions.

[WHT]

Could be just getting hidden node.Golly gee...sounds like UBNT needs to come out with a polling method.

[raytaylor]

Could be just getting hidden node.Golly gee...sounds like UBNT needs to come out with a polling method.

Polling. I always thought that was why you paid the extra cost to go for canopy or tranzeo?

Still, would be nice tho. Is polling a software thing or has it more to do with the hardware capabilities?

[WHT]

Polling. I always thought that was why you paid the extra cost to go for canopy or tranzeo? Well...UBNT has been on the cutting edge of more realistically priced hardware. You're not paying extra bucks just for the UBNT name.

[drwho17]

Could be just getting hidden node.Golly gee...sounds like UBNT needs to come out with a polling method.

Polling. I always thought that was why you paid the extra cost to go for canopy or tranzeo?

Still, would be nice tho. Is polling a software thing or has it more to do with the hardware capabilities?
Tranzeo doesn't poll, at least not their wifi, I've no idea why you pay extra for Tranzeo, they are atheros and their form factors such, their operating system is closed and in side by side comparison's the ubnt gear performs better at half the price. Must be the warranty.

Just a note, Nstreme was originally developed for backhauls, the Bullet-M series are advertised as backhauls, polling isn't that hard to implement, and they have some super secret feature for some reason people can't talk about, that only works with the new mode.

[WHT]

I've no idea why you pay extra for TranzeoYou pay for the name. An excellent example is a particular Cisco radome enclosed Yagi for $400 that has the same part number of the OEM's $100 antenna.

Some people had too many problems with single channel Nstreme. But its still a good solution when you use two radios at both ends.

However, the Bullet M will have similar speeds and it uses only one channel.

they have some (1) super secret feature for (2)some reason people can't talk about, that (3)only works with the new mode.That is correct on all three counts.

[Shockware1]

I think you are referring to TRANGO about the polling system.
Their 5830 AP and 5580 SU configurations use polling systems. 10 meg max per AP though, so they are gimped for speed.

For $2000 it costs for an AP I can setup an entire tower of UBNT gear complete with power conditioners, UPS, full grounding suites, ect... and still have money left over for redundancy. Probably get more clients hooked up per tower too.

And that's just one AP, You need at least 4 of them to be effective, 6 to provide full coverage while using up 100% of the available frequencies on both polarities.

It's a shame we wasted so much money on them in hind sight.

[lncommunications]

Bump: TPC and DFS?

Ty.

[UBNT-Mike.Ford]

Bump: TPC and DFS?

Ty.

Hello,

TPC and DFS are in fully supported in our new AirOS V

Thanks,

[Shockware1]

Bump: TPC and DFS?

Ty.
TPC and DFS are in fully supported in our new AirOS V



YEEEEEESSSSSS!

This allows us 3dB extra signal (legally speaking) to our EIRP yes? In Canada anyway...

Is it possible to port this functionality to other versions of airOS? or even better, have airOS V taking over 3.x.x?

TPC makes me happy :)

[lncommunications]

THANK YOU!!! You have made me very very happy! Thank you for listening to us :D

...and can these feature be implemented into the standard AirOS? it would be amazing if you could :shock:

[WHT]

YEEEEEESSSSSS!
THANK YOU!!! You have made me very very happy!You ain't seen nuthin' yet...More good stuff to come.

As usual, I can't discuss this yet.

[lncommunications]

Ooooo you tease! :P

[WHT]

At least ya didn't call me a "git" HAHAHAHAHHAH

[raytaylor]

Yes it is trango I was thinking about

DPC and DFS - tell me more????

[mobilexpert]

......
My main concern has to do with a limitation of 802.11a/g. Specifically, if a client on an AP is downloading at the APs full capacity, all other clients upload speeds drop to very low speeds. They jump around in the 0 - 200K range. Even worse, latency is severely impacted, which impacts VOIP.

I recently learned to resolve this with Mikrotik's newest version of nstream, but of course this makes CPEs much more expensive.

Soo...

2) Can nstreme work with the sr71 series?

3) Or better, does N resolve this problem natively?


about AP ....

every AP have own limitations in speed and client number,
usually limited by hardware or software.

example : mikrotik, on 2GHz AMD PC ... working in G-only mode, 20MHz channel, can deliver up to 25-26mbps data transfer HALF DUPLEX,

what it means ....

if 3 users are downloading at 5mbps each, and uploading at the same time,
they make 3x5 + 3x0.5mbps = 16.5mbps, and result is ... around 8mbps left

network without bandwith shaping for clients can easily go down on capacity, like situation you stated.
of course, i need more details to tell you what and how ...
(details of hardware used, exact number of clients etc.)

in reality, this setup gives very low lag, if LAN side is on quality backhaul,
or directly connected to router via LAN.

if you put 30 or more clients on single AP radio, with lower speeds each ...
it is also possible to work well, with some significant lag,
giving up to 10-15ms ping times under load - around 10mbps traffic going.

my personal opinion is that all is in hardware power of AP,
and of course using the same (atheros based) hardware (client cards or ap's)

there is one issue i found with mikrotik ....
it really need powerfull hardware, if you want powerfull performance.

of course, we use old pentium3 machines, Amd athlon old pc's etc,
each for AP, backhaul or 2-3 sites link etc, so the speeds over the air, with old technology go up to 50mbps for links up to 15-20 miles, 5GHz,
mikrotik of course, with nstreme + compression, backhaul links.

users are usually equipped with atheros cards, on 2.4Ghz and 5GHz,
but since 2.4Ghz became useless because of much noise,
we're slightly transferring all that is possible to 5GHz.

and regarding VOIP and other important services ...
you should use QoS, to make high priority to such services,
over the web, ftp, etc.

currently, NS5 and NS5L gives good results, and they also hawe traffic shaping option, which is required for our needs.
(for your needs - they have QoS also)

also, one interesting idea ... use 5-6 NS5 devices to cover 360 degrees radius, on AP sites, and sectorize users, so the single AP will never have more than 15-20 users, giving better radio performance.

b.r.
Alexandar

[WHT]

use 5-6 NS5 devices to cover 360 degrees radius, on AP sites, and sectorize users, so the single AP will never have more than 15-20 users, giving better radio performance. The Bullet M will change that model. Just wait and see....

[corky]

Mobile expert:
Thanks for the detailed reply.

In fact though, the problem we see is very different that what you describe. With 10mbits available bw, if 1 user is dloading unthrottled, uploads become very very slow, averaging < 200k, and occasionally halting for a second or two. Even if 1 user dloads throttled at 5mbits (we do sophisticated traffic shaping), uloads are dramatically impacted, the bw is not shared properly between up/down. This is true on towers with little cpu load on the AP. This is on a long range outdoor tower -- I doubt if this problem manefests where clients can "see" each other.

Nstreme's fixes all that. I'm wondering if N will have the same defect. If so, I won't be using bullets at all any more, but sr/xr boards in Mikrotik CPEs. In spite of the cost, the performance advantage is huge.

WHT -- do you have any insight into this?

[WHT]

I'm not sure of the question.

This is on a long range outdoor tower -- I doubt if this problem manefests where clients can "see" each other. Do you think you have a hidden node problem with the long range tower?

Nstreme's fixes all that.What problem is Nstream fixing and by what process?

[george]

Nstreme's fixes all that. I'm wondering if N will have the same defect. If so, I won't be using bullets at all any more, but sr/xr boards in Mikrotik CPEs. In spite of the cost, the performance advantage is huge.


That's the 64,000 dollar question of course. Its taken MT a long time to get Nstreme to the point where it works well, particularly regarding the low-latency-on-a-quiet-link problem. At a guess, Airmax may be some kind of polling analogue, but time will tell.

Without some kind of polling (or other method of allocating time slots reliably) you aren't going to get a large number of varied customers on a single AP. See Canopy, Trango, MT and others for reference.

The Ms should make for a super backhaul, but I reserve judgement on their suitability as an AP for PtMP until we see some real world results.

George

[Ron]

I'm wondering if N will have the same defect.

N does, but it looks like M has that problem solved ;)

[WHT]

N does what? What problem solved? How does "M" solve that?

[corky]

WHT,

With nstreme in it's latest incarnation, BW is shared properly between uploads and downloads, and latency is controlled adequately to be able to do voip and gaming much much more reliably, though still not perfectly.

George,
Is Airmax a future ubnt product/protocol?

Ron,
"M has that solved" ? M == n + AirOS introduces a protocol to resolve this problem?

[WHT]

Is Airmax a future ubnt product/protocol? No, its currently a working implementation, but we can't tell you at this point what it does or how it works.

[george]

No, its a future implementation as you can't buy it today. Really. All your bragging about how this is going to change the world don't make it so just yet... Although you've been kind enough to give us all a damn good idea what it achieves despite the NDA. Thanks for playing!!

George

[WHT]

No, its a future implementation as you can't buy it today. It is a current implementation and it works.
Whether or not you can buy the Bullet M from a distributor all depends when they get them in stock.

Although you've been kind enough to give us all a damn good idea what it achieves despite the NDA. Mike has ready given you an idea "what it achieves" (but not how MUCH it achieves, nor how it does it), and I merely backed him up. Ergo, no NDA problems.

Anything beyond that would be speculation on your part.

[WHT]

All your bragging about how this is going to change the world It won't change the world, unless you think a product that levels the playing field of a high end Canopy, SkyPilot, and (insert whatever other product you want) and for one tenth the cost... then it might change your world.

[corky]

Even though it's crazy frustrating to me as I'm making new deploymnets and eating up spectrum it will be hard to reclaim for something else, I understand why they are being vauge about this.

Keep in mind that Mikrotik took something like 4 years to get nstreme to work pretty well on point to multipoint. It is easy to get all excited when a product starts to *seem* to work, but there are many miles to go sometimes between that exciting day and a finished usable product a WISP could actually put customers on with confidence. Point to multi-point long range outdoor implementations in particular are hard to get right. Also, if they rush it, they might break things like future upgrade compatibility by not doing the design correctly.

I'm sure UBNT wants this thing released as badly as we do, and they will as soon as they have reasonable confidence in it. Until then, saying a lot would just create confusion and possibly undermine their credibility if it does not work out as hoped.

But it still drives me wild to put 100 units in the field in the next few months that may be instantly obsolete. Sigh.

[WHT]

But it still drives me wild to put 100 units in the field in the next few months that may be instantly obsolete. If you see a computer on the shelf at a computer store, its already obsolete. You can always wait for the latest model of smartphone/PDA to come out, but it will be obsolete in six months.

Nothing from UBNT is going to be obsolete...you will still be able to build out non-N 802.11a/b/g deployments with their current lineup.

So say you do finish or begin an 802.11g buildout, knowing the "M" will be a whole lot better. The situation is no different when 802.11g started to appear on the market with loads of 802.11b already in place.

I'm about to deploy Bullet M5 radios with less than a one mile radius footprint in an area that is saturated by WISPs still stuck with 802.11b and they are loading hundreds of users on a single AP (and suffering terrible complaints). I'll be offering twice the speed at half the cost. Will that make them sweat???

[george]

All your bragging about how this is going to change the world It won't change the world, unless you think a product that levels the playing field of a high end Canopy, SkyPilot, and (insert whatever other product you want) and for one tenth the cost... then it might change your world.

If it really does all that, it IS going to the be greatest thing since sliced bread. That would be very, very, cool.

I sound like a wet blanket on this because I've been personally involved in early deployments where problems didn't show up in beta, or indeed for some time afterwards.

Forgive my caution, but we replaced something like 60 SR9s in the middle of winter in -30* temperatures due to manufacturing defects. Ubiquiti (Mike Ford in particular) was great, but it still caused a huge pain in the buttocks and cost a hell of a lot of money.

Still, it shouldn't be too long now before we get something to play with.

George

[UBNT-Mike.Ford]

Hey Guys,

Streakwave got there Bullet M2 and M5's today.

Thanks,

Mike

[rconaway]

All your bragging about how this is going to change the world It won't change the world, unless you think a product that levels the playing field of a high end Canopy, SkyPilot, and (insert whatever other product you want) and for one tenth the cost... then it might change your world.

I will tell you that it is definitely a technological value but most of us play in pretty small fields. Until one of us wants to spend a million dollars doing a major city and demonstrating that it can compete with the big boys, it's a hard sell. I've agreed to fund a 4 mile deployment in Arizona so I get that demo site to work with. However, Firetide, Tropos, and BelAir have done a great job marketing to public safety. All of them have been hammered in muni-wireless so public safety is all they have left. We start going after that market tomorrow at a local Police Expo show.

On another note, Qwest, Verizon, and Commcast are spending millions on WiFi to offload the high-bandwidth apps on phones, so they see the value. They are also capable of funding these multi-million dollar projects, mostly with Cisco products. That market is drying up quickly. They still have a few years before they start looking more rural since they have 4G or WiMax products coming out to cover those areas.

The technology behind the Bullet M really changes the game. WiMax doesn't have a chance against the Bullet M or future products based on the same design, but the big boys are going to swamp the unlicensed frequencies in major cities. We have made some headroad into traffic and video but moving it back into metropolitan WiFi is going to take some serious funding or effort. We have a couple ideas on the table which we hope make that possible.

[WHT]

I don't fear WiMax. The services will be at least twice the price what I can offer. Cellphones work in non-LOS conditions because of the extremely low data speeds. It doesn't take a rocket scientist to figure what power levels are needs to deliver a 50 Mbps signal at 20 miles.

WiMax is like teenage sex...everyone is talking about it.

[drwho17]

I don't fear WiMax. The services will be at least twice the price what I can offer. Cellphones work in non-LOS conditions because of the extremely low data speeds. It doesn't take a rocket scientist to figure what power levels are needs to deliver a 50 Mbps signal at 20 miles.

WiMax is like teenage sex...everyone is talking about it.
Actually, you should try wimax before bagging it. Our initial field deployment of Alvarion is changing my mind, it's pretty amazing how it's working in 3.65. The QOS at the mac layer is pretty extensive as well, which is what we need, as we plan to do offer voice services to compete in previously territories where we don't have access to copper (Independent telco's).

I also would say, you should try a sprint wireless card sometime, in the deep woods in the far upper reaches of Michigan, NLOS I get 1Mb/s down in an area not supposed to be in their coverage area.

[WHT]

I have both AT&T and Sprint cards. The AT&T is like dial-up on diluted steroids, which ain't saying much. The Sprint card EVDO Rev A is around 750 Kbs to 1.2 Mbps, I've used it as a remote backhaul at times.

[rconaway]

Alvarion WiMax is very good. No question. However, it still can't compete against the Bullet M's head-to-head in performance, let alone compete against it in performance/value.

As for QOS, we are running Hi-Def cable TV video streaming across them now so voice isn't really a problem. We have seen an issue with channel changing with multiple cable boxes in the same house but that was early firmware. We haven't had a chance to put the latest firmware on there or finish making some adjustments (company put the program on hold for 3 weeks while they are sold to another company) but we will be back working on it in a couple more weeks.

This company considered the Alvarions but at $280 per client with $2K per base station made it a no brainer. Throw in twice as much bandwidth in the same 20MHz channel plus some future features I can't discuss, and there really isn't a comparison. Now, Ubiquiti isn't playing in the 3.65Gz band yet so that comparison won't happen. In the 5.8GHz, the Bullet's make the Alvarions not the better value.

[drwho17]

Alvarion WiMax is very good. No question. However, it still can't compete against the Bullet M's head-to-head in performance, let alone compete against it in performance/value.
Really? You are using the BulletM's in production, I think this is just wishful thinking, especially when BulletM's lack diversity, no 2nd order/4th order. I don't have any BulletM's yet, but I do have Alvarion's, it's a whole different level of wireless service.

As for QOS, we are running Hi-Def cable TV video streaming across them now so voice isn't really a problem. We have seen an issue with channel changing with multiple cable boxes in the same house but that was early firmware. We haven't had a chance to put the latest firmware on there or finish making some adjustments (company put the program on hold for 3 weeks while they are sold to another company) but we will be back working on it in a couple more weeks.

Voice is two way, and can't handle the jitters, multicast TV would be possible, although without polling you would suffer frequent rebuffers. Video falls apart, with any kind of utilization on a Ubiquity AP, we've piloted TV testing, it won't scale up to any amount of users through any non-polling wifi tech.

This company considered the Alvarions but at $280 per client with $2K per base station made it a no brainer. Throw in twice as much bandwidth in the same 20MHz channel plus some future features I can't discuss, and there really isn't a comparison. Now, Ubiquiti isn't playing in the 3.65Gz band yet so that comparison won't happen. In the 5.8GHz, the Bullet's make the Alvarions not the better value.
It's not the band, and those prices that you mention aren't really close for Alvarion's wimax gear, the CPE's are less, and the basestations are way more. I have it and the Ubiquities, I know exactly what they cost, since I deal with two of the 3 biggest Alvarion resellers, I'm not sure where you came up with those prices for BreezeMax. Alvarions for one, don't use 20mhz channels, 3.5/5 mhz is what you get, even at 3.65 they have 900mhz type NLOS, the performance at distance is much higher, they support 2nd order and 4th order diversity, I'm only using 2nd order in testing, 4th order would be even better, the base stations can handle 512 clients all with a guaranteed latency.

I've got an advantage on you, I have both gear types and can compare side by side, Alvarion's only downside is it's bandwidth limitations 6Mb/s x 6Mb/s (real tcp throughput), however we use 10mhz widths with 2.4, and throughput rarely goes to that, except with the best of signals.

[rconaway]

Keep in mind these are Bullet M's, not the current Bullet 5's. so I don't think you have had a chance to do a head-to-head comparison. Although the firmware is more mature on the Alvarion, the Bullet M's will be able to compete easily.

The Bullet M's are a single stream device but I suspect that it may not be the last of products coming out. Even with a single stream, the N's will exceed the throughput of the Alvarion. Voice is kind of a no-brainer with the type of features and performance in the Bullet M. If Ubiquiti comes out with a multi-stream 802.11N product with similiar features, this argument becomes moot. It would deliver 100Mbps in a 20MHz wide channel. As for your jitter problem, if you engineer the system properly in a fixed environment, this shouldn't be an issue.

The video streaming we are doing now isn't using multicasting. We are doing it straight through.

The prices I got are from our cable client who made the decision based on conversations with Alvarion directly. We are an Alvarion dealer but I didn't look up the price. I'll take your word for it. However, I believe it supports my argument. A tower full of Bullet M's may not be theoretically capable of supporting as many clients but it will support an order of magnitude more bandwidth capacity than the Alvarion and do it for 1/10 the cost. Without giving away any secrets, lets just say we are planning for several times more client capacity than current products per tower.

I don't think the Bullet M's can handle 500 clients but if you only have a 6MHz channel, I"m not seeing the value here. I do know that under full load, 100Mbps in a 40MHz wide channel, we saw the processors running 65%. That would be opposed to the current radios that would hit 100% at 33Mbps. Combine that features that cannot be named, I say the Bullet will have similiar guaranteed response times under load.

As for NLOS, you got me there. 802.11N handles it better than 802.11a and I suspect future products will be even better, that is where WiMax has an advantage. However, 802.11N has an ace in the hole on this, MIMO, which can support multiple polarities simultaneously. That sort of nullifies the multiple diversity advantage. It will be interesting to see which one ends up with more deployments.

[Ron]

we've piloted TV testing, it won't scale up to any amount of users through any non-polling wifi tech.

But you don't know whether or not the M does polling ... :)

[rconaway]

Not unless Ubiquiti tells me it does. You will have to ask Mike.

[WHT]

Mike said UBNT will have a press release by this weekend.

[jp498]

We've been using Alvarion since it was Breezecom and they made indoor industrial radios that a handful of people like us started using outside.

We buy an equal mix (quantity wise) of Alvarion, Trango, and MT/UBNT.

We just got a $60k grant project and we chose Alvarion for 90% of the radio system. They've currently got a real nice 900mhz system. Their 5.8 VL gear is very reliable.

Whatever their specs are, they are really conservative. If they say 32mb throughput, they mean 33mbps of real internet traffic when properly setup. They've had the advantages of Nstreme long before Nstreme was stable. Firmware comes out infrequently, and it is pretty good when it comes out. Documentation is fantastic with 500 page manuals in the form of searchable PDFs.

A big or growing ISP has at least four needs for radio systems. Reliable gear to prevent needless tinkering and service calls, high capacity, efficient use of spectrum, and inexpensive CPE. Alvarion is good at all but the last, though there are companies more expensive.

We only have a limited staff to install a certain number of customers every month in a timely manner, respond to service calls, build new infrastructure, etc... Tinkering and service calls prevent that growth and stagnate our progress of transitioning of customers to broadband (or better broadband)

Alvarion's got the best capacity at the moment for 900, and respectable capacity for 5ghz. As N and Mimo and such become more popular, I'm sure they'll have something for that.

Spectrum use; we have had 8 Alvarion systems in the 5.8 band at one site; that's some overlap. They also make stuff for other bands like the big 5.4 band, and we've even got some pre-ban 5.3 links still going. A big or growing ISP will run out of spectrum pretty easily, and UBNT can't do much legally outside of the 2.4/5.8 commodity bands to ease that. Alvarion has 3.65 ptmp gear, UBNT only has ptp oriented cards in that band (and those cards are working fine). We've run out of spectrum a long time ago, and every single AP we install needs planning to prevent interference and manage spectrum reuse, both with regard to our own systems and other local WISP use. I see some licensed 11ghz, 3.65, ptp links in my future and 5.4ghz legal ptp and ptmp gear needs also.

The Alvarion (and trango) CPE is more money, but the service calls are fewer, so it's worth while. Mixing in a little MT/UBNT where we know it will work well helps bring down average installation costs, especially for really small sites. No single vendor does everything ideally in all bands to meet all needs. It requires lots of testing of the various choices on the market and understanding your own needs.

[rconaway]

That's a very good analysis jp. Alvarion makes some of the best quality equipment in the industry. I have seen it make connections where other equipment sort of faultered.

In the 900MHz region, it's definitely a very good product. It would be nice if it had more power but no complaints. It's also faster than Motorola but Motorola will work with a higher noise factor.

I think the rate of service calls between Ubiquiti and Alvarion is about the same. I haven't had a service call on any Ubiquiti radio that was installed. However, you definitely have to take more care to set it up at the client to avoid these issue.

Historically, I also agree tha the Alvarion equipment will support more clients. The Bullet M's will change that though, and then you have to decide what you want to spend for installation. However, if you are already deepply embedded with client density issues already, tough call. You would be replacing a lot of equipment.

[masked]

\UBNT only has ptp oriented cards in that band (and those cards are working fine).

ubnt do have the nanostation3 but i believe it's only available for export?

[mobilexpi]

Yes, It's available only outside the US. Because it's not legal for sale in the US.

[UBNT-Mike.Ford]

The XR3-3.7 can be used as PTP or PTMP.

Thanks,

Mike

[rkj]

Have anyone noticed if Bullet M links at Gigabit with a Gigabit switch ?
Any word on VLANs on the config screens, or thru CLI vconfig ?

[rconaway]

The Bullet M has a 10/100 port on it.

[rkj]

The Bullet M has a 10/100 port on it.

Which makes me wonder how a Bullet M test achieved 104 Mbps on 100 Mbps limited device...

[UBNT-Mike.Ford]

The Bullet M has a 10/100 port on it.

Which makes me wonder how a Bullet M test achieved 104 Mbps on 100 Mbps limited device...

Wireless is half duplex not full duplex. 10/100 ports are full duplex.

www.google.com

Thanks,

Mike

[rkj]

The Bullet M has a 10/100 port on it.

Which makes me wonder how a Bullet M test achieved 104 Mbps on 100 Mbps limited device...

Wireless is half duplex not full duplex. 10/100 ports are full duplex.

www.google.com

Thanks,

Mike

TCP is a bidirectional protocol, even when it has to run over a half-duplex link, either 802.11 network or 10BASE-x network...

[WHT]

TCP is a bidirectional protocolTrue, but that has no bearing on what Mike is saying.

[UBNT-Mike.Ford]

TCP is a bidirectional protocolTrue, but that has no bearing on what Mike is saying.

This is correct.

802.11a/b/g/n is half-duplex regardless of the protocol going over it. Its just the nature of the hardware.

Thanks,

Mike