Pondering the future...

[Mark]

I'm sitting here at this moment, pondering which way to go for the future. As much as I've used star-os, as much as it's been my SOLE routing and AP system, I'm almost to the point where I can't anymore.

Let me explain... We gave up on Star-OS based CPE... for one, the cost. Not a complete killer, but nobody who ships quickly and is less than 3 days away has sold WAR1's in a year - and that's been the killer. Streakwave was excellent prices and best of all, they were very good at stocking and shipping when you wanted something. Nobody else comes close. The next closest just ships whenever the spirit moves them. Or, I can pay enormous shipping amounts...

Now, I am facing bandwidth demands that are simply overwhelming my ability to meet them. 33 2 meg clients on a 20 mhz channel sector and it's swamped. No, no real bandwidth complaints on it, but I can definitely see that we've reached the limits there. None of my Star-OS based backhauls will move more than 2000KB of "real" data. That means I have to use something else... and I'm doing that now. Can't wait.

So, as I sit here pondering what to put up next, as much of my system is current at max capacity, I'm looking at the brand X MIMO stuff. Going to order some shortly to test out.

AP plus antenna is only $230 for excellent gain and pattern antennas. 75M up and 75m down capacity... Clients are ALL under $100 including shipping, some under $50, with excellent packaging and form factors.

My business is now a partnership, and my partner's looking at the dollars and the issues we've faced lately and is going... "Why are we sticking with this?"

So, I'm reading the tea leaves... New MIMO star-os... Unless you're going to design and sell integrated CPE for 89 bucks... We're in a world of hurt.

The least expensive MIMO antenna for 5 ghz - that is a dual polarity panel 20 db arc wireless for about 80 bucks, add the required enclosure to fit the system and we're now up to 118. Presumably 2nd Gen WAR1 will be $70 + a radio of $40, pigtail, poe... That's about $180. Then you gotta get a mount and cable, etc.

Need a little more punch, 'cause you're not real close? 24 db MIMO panel and enclosure is about 160, which runs you up to about $230 for a CPE.

Find MIMO sectors other than UBNT's? Where? I can't find any, not in prices even faintly similar to theirs.

I was getting hurt big time due to the fact I had to have nearly $200 up front to hook someone up. DSL and Cable run ads promising to do it for free. So, I lowered my install and ate the cost of the hardware. Now I'm eating WAR1's at around 1 a month as they go belly up. I've eaten the cost of star-os CPE twice now, way too many times.

So, where are we going with this train? Is its destination even remotely sensible? Way back when, when I wanted a single board to do all, it worked and it worked for us for 4.5 years, worked awesomely well. But we're now lagging behind needs, mostly due to lack of capacity, and we're totally uncompetitive price-wise. Our service is lagging due to our need to expand many things and I have GOT to have some good reason to stick where I am... Because there's some darn good reasons NOT to.

Lonnie, you need to address where you're going with this, and quickly. Your customers NEED to know, and you need to offer something viable.

[riverdeltawireless]

I could'nt agree with you more. we just started a major system up grade last month. we also have a hodgepodge of equipment and its time to rain it in. its getting more difficult to compete with DSL and Cable.
Lonnie. i think i can speak for a lot of us here. we need info on where startOs is going
before we get to the edge. if i have to jump off i would like to know whats at the bottom instead of jumping blindfolded

[SpecialK]

Let me add my perspective to this.

1) Your brand "X" which you should really be calling brand "U" has lots of issues. Whey you have deployed their AP and have 30 subs on it then let us know how it is working, or put up a PtP link anywhere around other gear and tell me how you get through it, or stick up 3 of the brand "U" APs on the same tower next to each other and tell me how you do.

Like you guys I am in the same boat, customer bandwidth demands and competition is getting really crazy. If you think the answer is a $50 cpe then I think you have your choice already figured out. No one has a < $100 cpe but brand "U" and for good reason. Have you seen they have had a few recalls on defective hardware both APs and CPE? Brand "U" has no magic bullet unless they decide to make better gear that is more reliable.
When the noise floor really gets crazy you think all those CPE in plastic with a 45 degree antenna pattern are going to do any good?

2) More capacity is needed. 2.4 and 5.8 are getting slammed now and MIMO is only going to add to the problem.
WIth 20 and 40 mhz mimo antennas you are only going to get 2 or maybe 3 APs in the same space and not give yourself and room for self interference. We are going to have to move to 5.3/5.4 for short links and close up customers. Or even switch to 3650 quasi licensed bands.
But even that has little spectrum.


3) We are never going to be able to compete with DSL or cable.
You can not break the law of physics. In the end fiber is king and wireless will be used as alternatives to customers for PTP or used in areas underserved by fiber markets.


Even after saying this I am frustrated too.
I have been eyeing other gear as well and may be forced to try some real soon.
However pushing Lonnie into a corner asking for a magic bullet is just insane.
It helps no one.
I would say try the other stuff on a sector.
If it works go for it.
I would think there is no way you will see a < $100 MIMO CPE from Star-OS.

[lonnie]

11n is a speed increase, not a cure all. The new hardware and firmware is work in progress, and nothing else is getting in the way. It's coming soon, and if you cannot wait, then sorry, but there is nothing I can do to help you.

[Mark]

Let me add my perspective to this.

1) Your brand "X" which you should really be calling brand "U" has lots of issues. Whey you have deployed their AP and have 30 subs on it then let us know how it is working, or put up a PtP link anywhere around other gear and tell me how you get through it, or stick up 3 of the brand "U" APs on the same tower next to each other and tell me how you do.

Like you guys I am in the same boat, customer bandwidth demands and competition is getting really crazy. If you think the answer is a $50 cpe then I think you have your choice already figured out. No one has a < $100 cpe but brand "U" and for good reason. Have you seen they have had a few recalls on defective hardware both APs and CPE? Brand "U" has no magic bullet unless they decide to make better gear that is more reliable.
When the noise floor really gets crazy you think all those CPE in plastic with a 45 degree antenna pattern are going to do any good?



The thing about it is, their CPE are NOT 45 degree antennas. They're 22, 21, and 27 db antennas. There is that little short range "loco" thing, but it hardly qualifies as useable in my area, since I do so little short range stuff. My customers are generally between the 1 and 14 mile range. Mostly in the 4 to 9... I never used any of the Star-OS integrated CPE precisely because of the low gain antennas and wide patterns. The lowest gain 5 gig stuff I've ever used was 21 db grid. Mostly, it's been 25's and 23 panels.


2) More capacity is needed. 2.4 and 5.8 are getting slammed now and MIMO is only going to add to the problem.


Yeah, fundamentally agreed among all of us, I'm sure. However, a MIMO system gets you about 3 to 4 X the amount of data to customer for frequency used - ergo it DOES conserve RF real estate. My one sector that serves 33 on ONE frequency should be able to serve 60 - 80 in the future, and still have considerable reserve bandwidth. Another thing we're doing is trying to go to smaller cells. Smaller areas with geographic curtains from elsewhere, etc.

WIth 20 and 40 mhz mimo antennas you are only going to get 2 or maybe 3 APs in the same space and not give yourself and room for self interference. We are going to have to move to 5.3/5.4 for short links and close up customers. Or even switch to 3650 quasi licensed bands.
But even that has little spectrum.


3) We are never going to be able to compete with DSL or cable.
You can not break the law of physics. In the end fiber is king and wireless will be used as alternatives to customers for PTP or used in areas underserved by fiber markets.



I have no need to, nor will I probably EVER compete with fiber, as there's unlikely to ever be any around here. Still, I have to be able to move the bits to the consumers, in ever larger amounts. 2 years ago, people were impressed with 2 meg. Now they're like "what's the next step up?" and I don't have one, and can't deliver one. I've actually gotten away with this longer than I expected to. I had originally predicted my needs would swamp this system 2 years ago, at 3.5 years, not 6 months ago. We're all living on borrowed time right now. Tain't Lonnie's fault or anyone else's, it's just that that's what's going on and we need to find our means of dealing with it.


Even after saying this I am frustrated too.
I have been eyeing other gear as well and may be forced to try some real soon.
However pushing Lonnie into a corner asking for a magic bullet is just insane.
It helps no one.
I would say try the other stuff on a sector.
If it works go for it.
I would think there is no way you will see a < $100 MIMO CPE from Star-OS.

It doesn't HAVE to be under a $100. But over $200 just isn't going to cut it anymore. Not unless we have some magical bullet that can deliver 300 MBit per sector would it be worth investing in at over $200/customer.

My bandwidth consumption is growing somewhere between 3 and 8 percent PER MONTH. Video stores are closing and I'm seeing movie downloads 24/7 and TV show watching growing fantastically. The name of the game is deliver... Or lose it. I'm not one to give up, so...

[Mark]

11n is a speed increase, not a cure all. The new hardware and firmware is work in progress, and nothing else is getting in the way. It's coming soon, and if you cannot wait, then sorry, but there is nothing I can do to help you.

I don't think you got my point, then.

The absolute finest RF delivery system in the world isn't going to help us, if the availability of the rest of the parts isn't a workable option. Star-OS as a unique system was very much so when it started, but not anymore. Your competition has continued to add features to their software, so many devices now can route, NAT, dhcp server, etc, etc.

Let's take UBNT... They have lesser capable RF and routing (it almost doesn't exist), but they chose to divide their attention between cost / form factor and rf capability.

You've made the 11a/b/g system perform about as good as it can, but that's all you've addressed - in the past, that was BY FAR the biggest factor - my old Star-OS 11b sectors are still humming away with 15 to 25 clients and they're STILL performing very well, considering thier amazingly lower theoretical data rate. I'm still and others are too... Impressed that I can easily deliver 2meg to a customer at 18, 21, 23, 25, 29 miles without difficulty.

But the market and technology has evolved around us.

There once was a point when almost nobody addressed the cost of antennas, enclosure, etc. The integrated systems cost more than what we were building. Alvarion, Canopy, etc, are still around, but still expensive. Many people did all kinds of competitive stuff concerning enclosures and the little parts - all of which have fallen dramatically in price from their 10 year ago highs.

Now, it's being done, and just focusing on ONE aspect, when your competition is focusing on 3, isn't going to keep US in business.. Nor you.

If you can't find a path, using your products, to a solution that's competitive and workable, and explain to us how it works in the future, given the state of the market, then you've failed to plan - which is something that this guy named Lonnie used to harp constantly at us about...

"Failure to plan is planning to fail". - Lonnie N

Me and Special K, at least are reading your words, and reading between the lines. Other people have addressed MIMO, and have a half year of bug squashing and fault finding by the marketplace already, and we've not even gotten started. I'm still waiting for some kind of definitive wrap up to the last re-do of V3, which has been a LOOOONNNGGG time coming.

Do you have a plan to keep your customers competitive? Have you addressed, in ANY fashion, how evolution has occurred since you got out of the business of being a WISP for your own business model?

If so, we, your customers are now asking you for some frank input.

[Mark]

One final note...

The reason that producers like the former Trango, Alvarion, Motorola, and so on, kept building "integrated" solutions was often to keep the prices UP, and to keep the economies of scale as profit for themselves. And, as an aside, quality control.

There have been many integrators to come along since then, who package bits and use economies of scale from different sources to address price - And many of us could name a handful of them.

Now, it's evolved beyond that. Now it's "economies of scale" used to integrate commodity technology into an "economies of scale" priced devices, and it's being done now to OUR advantage.

Whether I LIKE it or not, my biggest competitor is now going to that advantage. We used to be on the same foot, they used that Latvian mess, but the rest of the stuff was the same, now they're moving to UBNT, and can offer pretty much minimal charge installs and they're going to make a profit, even after all costs, in 90 days.

And I've got a wait 9 months to a year to break even?

This is NOT that hard to understand. It is NOT a competitive business model. Even if we have to change 10% of our installed gear every year, we're still far ahead on viability - and sadly, my star-os gear is failing at half that rate now. Especially WAR-1's.

You built a competitive business model when the competition was just about performance and features. And it worked for a long time. Now the market's changed. How are you changing, and how will it work for us?

That's what has to be answered.

[bobbyc]

I invite you to try the ubnt system. I live in a pretty rural area, and finding a clear 2.4 or 5.8 channel is hard enough. When you are dealing with dual polarity antennas, it's near impossible around here to make it work. The ubnt single chain stuff (bullets) are hard enough to make work when there's noise up on a tower. 1.3.23b (b and g) just kick its ass. Way better thruput.

IMO this MIMO/dual chain N stuff is not going to fly outdoors if there's ANY noise present... hopefully the starOS driver will be way better. I know better to use dual chains/polarities for one sector/system though. I'd rather use that clean channel on another side of the tower. Single chain for me!
BOb C

[lonnie]

Look, it has been shown that looks good and cheap are not really very good to base decisions on for wireless. It has to work good -period- so we really do not care if the competition is addressing looks and cost. We will focus on performance.

You decide, but don't be blaming me for any poor choice you make.

Many companies have made HUGE profits selling cheap and/or flashy junk. We are not one of them, which is why we are sometimes late to the party. Once we arrive we always set the standard to be judged by. Quality takes time. Some of you have not yet learned that.

[greg]

Let me explain... We gave up on Star-OS based CPE... for one, the cost. Not a complete killer, but nobody who ships quickly and is less than 3 days away has sold WAR1's in a year - and that's been the killer. Streakwave was excellent prices and best of all, they were very good at stocking and shipping when you wanted something. Nobody else comes close. The next closest just ships whenever the spirit moves them. Or, I can pay enormous shipping amounts...
I've been buying mine directly from George on the Star-os web site. Almost always they are shipped the same day of the order. I get my shipments in 2 days and pay ground shipping. I had been specifying Fedex since UPS delivers so late in the day. There have been times when they're out so I do tend to keep some in stock and order sooner. The shipping used to be pretty high when coming from Canada but I have no issues with it since they moved operations to Minnesota.

I have accumulated quite a few broken Tranzeos and have been installing war1 boards in those antennas after finally finding and getting the right pigtails. They are a nice combination.

[Mark]

Look, it has been shown that looks good and cheap are not really very good to base decisions on for wireless. It has to work good -period- so we really do not care if the competition is addressing looks and cost. We will focus on performance.

You decide, but don't be blaming me for any poor choice you make.

Many companies have made HUGE profits selling cheap and/or flashy junk. We are not one of them, which is why we are sometimes late to the party. Once we arrive we always set the standard to be judged by. Quality takes time. Some of you have not yet learned that.

Am I to assume that the answer is "No, we have no intention of addressing cost and equipment availability, or even have any idea if what we're embarking upon is viable as a business model"?

Let's see... I have no idea where I said that I wanted "cheap flashy stuff" that "looked nice". I read and re-read what I said ,and it's just not there. I'd appreciate if you did not put words in my mouth, especially when the topic is as serious as this.

But, it hardly seems like you should be going on about quality... I have a LARGE stack of dead compex mini-pci radios. MOST of the WAR-2 boards I have are now dead. The number of WAR-1 boards I bought and are now decorating the dead board pile continues to climb. I've managed to throw away EVERY pigtail I ever bought from Valemount, because they failed or weren't any good to begin with.

But, none of these did you engineer yourself, and the QC of Compex is not fully under your control. Notice I haven't posted long rants in the forum blaming you for these failures. I gave you the benefit of the doubt that you did your part of the bargain and consider my losses as part of the cost of doing business. But to tell me that somehow, Valemount's stuff is the premium of the lot, and that brand X and Y, after having produced stuff, debugged and steadily improved stuff over the years, are "cheap flashy junk" in comparison is NOT valid, from any standpoint.

Every vendor out there has had teething issues for new products - just like we have a long series of betas between each release. All of us expect that in some way, we're going to pay for being on that edge.

None of that is or was the point of my posting. The point of my posting is, WHERE ARE YOU HEADED? And is what your producing going to be a viable business model? This is nothing to do with mischaracterizing OTHER people's stuff. This is asking YOU where YOU are going. Flung mud at other people is no substitute.

[lonnie]

Mark, I have said we are moving to 11n and I have said that it is work in progress with nothing else getting in the way.

Our gear is not over priced and it works. If you want something cheaper then you know what you have to do.

We are developing for the ISP and of course I know it has to be affordable. Will we meet the cheapest price? I can say no, we will not, with the line of gear we are going to use.

How can I answer whether our units will be a viable business model? First off, the market will decide that, and second, a public forums is certainly not the place I will discuss our business.

To sum up, we are close to having 11n hardware and firmware. It will be a bit more expensive than 11abg gear due to radio costs and antennas, but it will be affordable. It will have higher performance but will require careful deployment.

[go.fast]

Are you going to have complete assembled units, including antennas? Just hang the unit, plug the cable in and go?

[go.fast]

I too would tend to agree with quality issues.
Power supplies have been mediocre to horendous. So much so I usually buy my own and don't buy your kits. Why waste my money on a PS that's going to make have a failure and truck roll as well as the most important thing, a dissatisfied customer.

Another thing that I did not like to much is to see the same antenna unit on 2 gig and 5 gig units.

And we never did get an XR-9 to work on 1.4.x ..

Hope the N is better than what we have had up till now. Won't be any good if you can't have simultaneous users downloading and uploading megs of traffic..




Am I to assume that the answer is "No, we have no intention of addressing cost and equipment availability, or even have any idea if what we're embarking upon is viable as a business model"?

Let's see... I have no idea where I said that I wanted "cheap flashy stuff" that "looked nice". I read and re-read what I said ,and it's just not there. I'd appreciate if you did not put words in my mouth, especially when the topic is as serious as this.

But, it hardly seems like you should be going on about quality... I have a LARGE stack of dead compex mini-pci radios. MOST of the WAR-2 boards I have are now dead. The number of WAR-1 boards I bought and are now decorating the dead board pile continues to climb. I've managed to throw away EVERY pigtail I ever bought from Valemount, because they failed or weren't any good to begin with.

But, none of these did you engineer yourself, and the QC of Compex is not fully under your control. Notice I haven't posted long rants in the forum blaming you for these failures. I gave you the benefit of the doubt that you did your part of the bargain and consider my losses as part of the cost of doing business. But to tell me that somehow, Valemount's stuff is the premium of the lot, and that brand X and Y, after having produced stuff, debugged and steadily improved stuff over the years, are "cheap flashy junk" in comparison is NOT valid, from any standpoint.

Every vendor out there has had teething issues for new products - just like we have a long series of betas between each release. All of us expect that in some way, we're going to pay for being on that edge.

None of that is or was the point of my posting. The point of my posting is, WHERE ARE YOU HEADED? And is what your producing going to be a viable business model? This is nothing to do with mischaracterizing OTHER people's stuff. This is asking YOU where YOU are going. Flung mud at other people is no substitute.

[Mark]

Are you going to have complete assembled units, including antennas? Just hang the unit, plug the cable in and go?

I dunno that we need all that, but man, it would be nice if they could get at least a 22 and a 27 db dual pol dish certified, or even better, panels - something that doesn't cost a fortune.

UBNT recognized the enormous and absurdly high cost of dual pol antennas - along with their scarcity and arranged to have some built at "commodity" prices. They integrated the previous physical form directly into commodity antennas, and built wholly integrated cpe. The reason? Lowering costs, so people buy more.

They've even got solid dish CPE for under $100. Now, I'm not thinking I have to have under $100 to survive. I surely don't. But $300 for a cpe? I might as well give up and go flip burgers. There's WIMAX cpe for well less than that, for pity's sakes - yea, LICENSED WIMAX cpe.

Once you past the custom PC board inside the Bullet case, the shell on the outside is just commodity stuff. So are grids, etc.

There's nothing that says that Valemount can't have any one of 300 desperate for a product to build and sell Asian makers put commodity stuff together.

Sure, you can sell us bare boards. We'll need them, too. Just as I use them now. But we have to address how the world has transformed since the last version of STar-OS and the WAR boards came into being. I'm quite sure that they could find ALL the different components are actually now commodity bits, from dipole designs to mounts, etc.

That's why multiple brands of antennas have exactly the same mounts, now.

Yes, we'll need those high end dual pol antennas and stuff now and then. But, we can buy them, too. We just need good daily prices for commodity parts now.

[SpecialK]

Mark,

Do you have any brand "U" gear installed yet?
I bet not?

Let me rant for a bit.

First off, you say it's not about price yet right in your original post you say that if there is not a Star-OS mimo device for $89 you are in a world of hurt.
And you mention the $100 22 db solid dish brand "U" device. It's not even out yet so how can you talk about it.

#2, Brand "U" I believe uses pac wireless for the antennas. I do not know about you but my little experince with that brand of antenna is not so good. I come from the wireless world of using Til-Tek,Radio Waves, and MTi for sectors, panels, and dishes and there is no comparison in quality and performance. Every sector I have up with one of these high end antennas is working very well even in a 2.4 AP that can hear no less than 30 home routers.

#3, lets talk Wimax for just a moment. What band are you going to run in? 3650 is about it for us unless you somehow have a licensed frequency which I doubt or you would not be looking for a $89 cpe. Also 3650 has a wopping 50 mhz of space for us and 1/2 of that is not usable to most of the vendor equipment because of certification. Have any competition around you? What if they also deploy 3650, are you ready for that? Not to mention if there are 2 or 3 competitors around you at all. Sure you can get a bunch of pipe thru a 5 or 7 mhz wimax channel but just because it is in the 3650 band does not make it a true Wimax system.

#4, It seems you have it all figured out. You source from Asia using all the cheap off the shelf parts and make the best $89 cpe in the world. I say go for it. Where is your R&D? How many different antennas are you going to purchase just for testing? Not to mention coding radio drivers unless you are just going to take openwrt and slapping on the board. If that was so easy and worked so awsome you could be buying from all the vendors already. The simple truth is right now they all have issues.

#5, Lets hear from some of the guys on here that have deployed brand "U". There are plenty of them posting on the brand "U" forums.
Not giving any vendor specific names or praises but lets here if any of the guys deploying brand "U" will stake their entire WISP business on it right now? This will give us insight into if a $89 cpe available right now today is the greatest thing since sliced bread.

Now for my flip of the coin.
I understand you are frustrated right now. So am I.
I have not seen a fully functional Star-OS software version that does not have bugs or that contains all the normal features that any WISP based wireless product should have like WPA, WPA2, and bridging that truely works. I'm not sure why it seems like V3 1.3 and 1.4 are abandoned. At least the bugs could have been fixed or normal features like WPA could have been added. I too am deperately hoping Lonnie's next version of whatever it is, addresses our needs, mainly in price/performance, form factor, and feature set and soon. It may be my last attempt at using Star-OS myself. I also wish Lonnie would give us some more info. I really think he is staking his entire current business on this next release. I think there are plenty of ways to inform of us what is going on better without out giving out the family trade secrets. But that's his tight rope to walk not mine. Leaving us with so much doubt about the current versions and future versions is also giving us the reasons to pursue other vendors as well. Hopefully Lonnie can step back for a moment and see it from our perspective as well and see that it looks very different from our side of the coin. It's been over 2 years since we had a version of Star-OS with all the features we need in it. It's been over 7 months since we have had a version of software that is even usable, not to mention contains all of the features we need.

[lonnie]

I've said before, but it is worth repeating.

Things have gotten way more complex over time. In the old days Tony could bang out a new driver in a few months. The source code base was 200K lines. Now we are dealing with drivers of a million lines of code and more, and obviously several teams have been working and fixing/enhancing. Some of the code is slick, some of it is very broken. Our job, as always is to fix the broken things and add some useful enhancements.

We are not first to the party. We are aware that others might already be to market, but maybe they rushed a wee bit. It gives the impression they are leaders, but some people are fooled.

I do not wish to give much more than this. If others know what we are working on, they will work on it too. Not good in today's market, at least for us. We will be to market soon, and it is the only thing we are working on. I cannot give a time estimate because this is software, and it can be like the mythical 10 minute job.

Again --> we are close, so your patience is appreciated.

[riverdeltawireless]

Let me chime in on this a bit

first off we started using ydi Crap when we first started and have tried every thing till we found staros. ubnt at the same time and for us star AP's and ubnt CPE has worked so well that it is rock solid even bridged (converting to routing in process)
the form factor of the ubnt product with the staros OS on it would ROCK
but as i understand it costs to design and build

[Mark]

Mark,

Do you have any brand "U" gear installed yet?
I bet not?

Let me rant for a bit.

First off, you say it's not about price yet right in your original post you say that if there is not a Star-OS mimo device for $89 you are in a world of hurt.
And you mention the $100 22 db solid dish brand "U" device. It's not even out yet so how can you talk about it.



Let me address the pricing issue. I was clear, was I not, that meeting an 89 dollar price wasn't necessary? But I was clear that raising the price of CPE NOW, to higher than Wi-MAX prices (near $300) isn't viable, especially since almost ALL Star-OS users are small businesses. Lonnie doesn't have to say where he's going or go into great detail. The problem lies in that we have absolutely NO direction at all apparent, and "if you can't wait, just move on" isn't reassuring at all. If in the future, I'm intending to use $170 CPE, then I'd better not be shuffling my prices and rates and fees to become competitive, because I'll just have to undo it. I'd also better prepare to grow much slower and lose more to my competition. Or, rethink my network and start looking at other vendors. You can buy 3.65 WiMax CPE for less than 240. Tempting, very tempting. The AP is pricey, though, be warned. Still, they use VERY small spectrum slices and what I do have for 3.65 in place has worked really, REALLY great, due to lack of noise.

I have a bunch of Bullets in place. As legacy A/B/G gear. Not wholly impressed with them, but they do work, and the failure rate isn't any greater than my star-os based stuff. The form factor gives fewer leak/failure points, so that's a positive.

I haven't put any of the Airmax stuff in production yet, but it should be in a few days. Just a P2P link at the moment, and then in a month or so, a 5 ghz sector on one of my lesser used sites, where I can test distance, performance, and how well it all stands up to noise, etc.

The noise handling of the UBNT stuff is definitely a step below most of the Star-OS stuff, but we've actually had it work where the Star-OS stuff didn't, so... hard to quantify real well, though the generalization seems to hold quite true. The ability to screw the device into the back of a grid or panel directly has been a VERY decided advantage. I can count on signal levels for 5 ghz stuff to be on the average of 3 to 7 db better than I otherwise expected.

Certainly, the 2.4 Bullets are mostly worthless for any distance, but the 5 ghz ones are near indistinguishable from star-OS in terms of performance, throughput, and handling distance issues. I have only had 2 field failures in a year, both the same place, both the same failure. I have had a higher than expected "not working when new" rate, but if you add up all the antenna, radio card, pigtail, power, enclosure, and other problems, the overall outcome's probably insignificantly different.

The highest rate of failure of ANYTHING has been PacWireless POE's. Lost a bunch of those, and the second largest has been WAR-1 boards...blown capacitors, sudden death, or instability and you find out they're making noise when you investigate. Single largest failure has been the ethernet ports on WAR-1's, though. I don't know how many WAR-2's I bought, but as of now, I have just ONE left functional. The rest self corrupted with bad firmware, or just went stone cold dead, or the ethernets died.

The ultra-cheap version of the WRAP board was a flop, as well. I had to get those back when there wasn't ANYTHING else due to shortages. I'm sure all of them are now gone. They died in all kinds of creative ways. Most WRAPs have died by no longer detecting the radio cards, or by ethernet failure. The visibly scorched ones (of anything) I considered to be outside factors :)

ALIX boards have a lot of CPU power, but seem to have relatively weak card power (use one high power XR card and you're treading on thin ice) and rather fragile ethernet ports, as well. Sure do like the ram, cpu horsepower, and bootup speed, though. Price is so-so.

I have multiple WAR-4's and no failures of any kind... or complaints.

[Mark]

Perhaps this is more relevant... But for the moment, if low cost CPE can't be done, then perhaps pursuing it is pointless.

I don't mind at least trying a "distribution and microcell" approach to dealing with the large bandwidth growth. But at the moment, we're just shooting ourselves in the foot to try to deploy greater numbers of legacy A/B/G stuff, as we're crowding ourselves out of spectrum and have no chance of serving the numbers of customers we're passing with our signal, as the quantity of throughput just isn't there.

I noticed a lot of commentary about using single chain N gear. I have no intention of doing so. My own opinion on customer demand is that we need to be prepared to maintain sustained 2/meg customer in 1-2 years. Since my customers are far apart, and geography puts many of them at 2 - 12 miles, the number I need to feed per antenna/spectrum slice is going to have to be relatively high in some places.

No, I don't mean 2M / customer 24/7, but 2m to at least half of them for 1-3 hours per day. We're going to see SERIOUS increases in the number of bits transmitted, far more than we have so far.

So, while my original 2M customer projections worked out fine for 5 years, I think my next step up is going to be 5 to 10 and that the viability of that speed is going to be less than 5 years long. I would say that currently, 10% of my customers use internet based entertainment - like netflix and streamed tv shows. I know without a doubt that number is bound to increase to at least 30 and possibly as high as 50% in the next 1-2 years.

So, yeah. My next step is be prepared to, if not actually do, deliver 5Mbit per customer, with the AP capable of delivering 100Meg.

Using my best guesses... That's 5/100 = 20 X 5x oversubscription - 100 customers per AP. Right now... I'm 2/15 = 7 X 5X oversub, that's 35 an AP and it's about maxed out. I've got 27 on a 10 mhz wide AP and it can't quite keep up.

Right now, I'm working on my guesses as to what to put where, future B/W needs to each, and how that works for physical space, blah blah. Thus, my need to decide things RIGHT NOW.

Since bandwidth use has changed so much in the last year, I think we're probably mostly in the same shoes, needing to plan future viability.

[luke541]

Mark, surely you have something better to do than spend your days and nights flaming the forums about Your needs, Demands and Desires. Why can’t you just calm down and wait as the rest of us will, and have too? We the customer(s) don’t Run Lonnie’s business and for them to release before they are confident they are ready to is foolish. With that said, I highly respect their decision to release when THEY are ready to do so, not when we demand it. If you can’t wait, then maybe you should migrate to that other brand. I do think it’s just awesome that an 89 dollar brand U AP can feed 300+ 89 dollar sub’s…. LOL, sorry just choked on my wine……. Totally makes sense though right?

With all that’s been said, this discussion should be put to a stop. The questions that Can/Will be answered, have been. What you and the rest of us all want will come, we all know it will. But for us users to put the timeline on when that will be is not going to get us a reliable stable product. I’d suggest you be patient. If patience does not pay for you, we’ll see you when you come back!

[jeff]

Mark, surely you have something better to do than spend your days and nights flaming the forums about Your needs, Demands and Desires.

Companies that listen to their customers and meet their needs continue to have customers.

Companies that dictate what their customers should need fail in the face of competition.

I saw no flaming in this thread and was quite impressed that Lonnie let it continue rather than delete the whole thread. Some of Mark's points are accurate and well said.

[Stratolinks]

I do think it’s just awesome that an 89 dollar brand U AP can feed 300+ 89 dollar sub’s…. LOL, sorry just choked on my wine……. Totally makes sense though right?

Lonnie has deleted some text.

After 8 years of putting WiFi equipment into harms way (on towers and grain elevators), I have lost 2 WRAP boards with radios, and 3 Omni antennas. Yes we have lost a few radio cards here and there, but not usually to lightning (at lest not coinciding with a thunderstorm anyway), they just failed over time.

EVERY WAR4-Metro I ever purchased is still working, Every WAR-2 board I ever purchased is still working. We have lost a few WAR1 boards in CPEs, but considering we don't put any lightning protection on them, we don't ground them, and we live in the lightning capital of Canada, we have done just fine. (We require the customer to provide power protection to them). Percentage wise teh WAR1 are more than 3 times as reliable as the previous gear we put up.

I will gladly wait a few months to see what the Star-OS crew brings to the plate before I make any major changes to my network.

[tog]

I... yeah basically everything that stratolinks just said right down to me also being in a lightning capital.

I must confess I put a couple of "other company's" units up as backhauls and guess what? I am climbing all three towers they are on to swap them out because 4 out of the 5 total units I bought have a manufacturing defect with the radio.

[lonnie]

This thread has really gotten away from the topic, and I want to stop it before it gets too crazy. Please no more good or bad reports of other gear.

Some very specific posts have either been deleted or edited.

[handyman]

Now for my flip of the coin.
I understand you are frustrated right now. So am I.
I have not seen a fully functional Star-OS software version that does not have bugs or that contains all the normal features that any WISP based wireless product should have like WPA, WPA2, and bridging that truely works. I'm not sure why it seems like V3 1.3 and 1.4 are abandoned. At least the bugs could have been fixed or normal features like WPA could have been added...

It's been over 2 years since we had a version of Star-OS with all the features we need in it. It's been over 7 months since we have had a version of software that is even usable, not to mention contains all of the features we need.

This summarizes my predicament regarding access points. I cannot use 1.4 because WEP has been repeatedly hacked on my network and I need WPA. Cannot use 1.3.23 because I rely on isc-dhcpd which is buggy in 1.3.23. It seems that 1.3.13 has the habit of letting radio cards get wedged requiring manual restarts too frequently. And 1.3.13 won't even load onto the newest 4-port boards. So I cannot use new boards. When lightning takes out some more APs this summer, what am I to replace them with? I have a dwindling supply of old boards with 2 ethernet ports that can run 1.3.13. I have to start testing some alternative this spring so going into lightning season I have a plan B.

The last few years have seen the steady disappearance of features I came to depend on: 2nd Ethernet port, dhcp service, WPA. This happened while the quality of the wireless performance was improving significantly. It seems that Valemount is getting the hard problems solved while ignoring the easy ones.

[Mark]

This summarizes my predicament regarding access points. I cannot use 1.4 because WEP has been repeatedly hacked on my network and I need WPA. Cannot use 1.3.23 because I rely on isc-dhcpd which is buggy in 1.3.23. It seems that 1.3.13 has the habit of letting radio cards get wedged requiring manual restarts too frequently. And 1.3.13 won't even load onto the newest 4-port boards. So I cannot use new boards. When lightning takes out some more APs this summer, what am I to replace them with? I have a dwindling supply of old boards with 2 ethernet ports that can run 1.3.13. I have to start testing some alternative this spring so going into lightning season I have a plan B.

The last few years have seen the steady disappearance of features I came to depend on: 2nd Ethernet port, dhcp service, WPA. This happened while the quality of the wireless performance was improving significantly. It seems that Valemount is getting the hard problems solved while ignoring the easy ones.

Valemount basically writes the drivers and interface, the rest is open source stuff that they're using. It's not their fault if ISC-DHCP is buggy, or if any of the routing daemons don't work right. They're not the authors, and learning those things and debugging them would be a huge additional workload.

I have no idea what percentage of time they're devoting to what - and this includes V2 and V3, etc, but trying to fix the open source stuff they integrate into their products probably isn't really an option.

Maybe in the future there'll be an option for us to compile and update binaries on our own and that would take some of the workload off them, as it concerns fixing things like the DHCP daemon. That way those of us who have the ability can quickly fix things like that, and the number of needed releases and compilations across all the platforms would be smaller.

Even just a tool to update binaries in image files, binaries we download and install ( must have a security hash file to integrate or something? ) into our images and then we update our systems... More work for us, but maybe less need for them to update versions.

[handyman]

Valemount basically writes the drivers and interface, the rest is open source stuff that they're using. It's not their fault if ISC-DHCP is buggy.

Indeed. That's why I would classify this as an "easy" problem for them to fix.

The bug was introduced in isc-dhcpd version 4.0.0 (Dec. 2007) and fixed in 4.0.1, Dec, 2008. Actually, 4.0.0 has rather a long list of fixed bugs. 4.1 has been out for more than two years now, so it just remains for Valemount to build a release of 1.3 using a release of isc-dhcpd that is less than two years out of date.

[Mark]

Indeed. That's why I would classify this as an "easy" problem for them to fix.

The bug was introduced in isc-dhcpd version 4.0.0 (Dec. 2007) and fixed in 4.0.1, Dec, 2008. Actually, 4.0.0 has rather a long list of fixed bugs. 4.1 has been out for more than two years now, so it just remains for Valemount to build a release of 1.3 using a release of isc-dhcpd that is less than two years out of date.

I have mainly used that version for nearly everything on my network. None of the bugs have been fatal to me, nor have I had any issues I know of I can trace to it. Maybe that's why it hasn't been updated. Not that big of an issue for most.

[handyman]

I have mainly used that version for nearly everything on my network. None of the bugs have been fatal to me, nor have I had any issues I know of I can trace to it. Maybe that's why it hasn't been updated. Not that big of an issue for most.

If you do not deliver DHCP service to customer routers (esp Netgear), then you will not see the bug.

[amp]

Well i don't want to buy another war1, my hot climate kills them fast. If they are in direct sunlight they get up to 100f even in winter. I am nearing 50% failure rate, the other day i replaced one customers unit the 2nd time in 18 months since the original install.

Having recapped a stack of boards the main common denominator was <ZL> caps. Bulging or leaking. From my research there is no known brand ZL. There is rubycon ZL which is the highest end capacitor brand out however those wouldn't die and they would write rubycon. I have only seen two black caps, chang and zl. Some boards would have a mix and only the zl would die. To have this many winter deaths is crazy.

I have taken down working boards with budging caps so I'm worried more are waiting to die all the way. My main truck roll is that now. Its hurting my perfect reputation.

I am glad to hear mikrotik support, a friend who uses them reported 1/80 cpe failure. Just hope it happens soon.

So i am not sure if i should recap every board with new or if the recent boards have caps that will last. I should of been keeping better track of the brands i was replacing. It could of just been that run a year and a half ago. Time will tell if all the new colored caps die too.

I only have one bad Ethernet war1, it randomly boots up and doesn't connect only when outdoors even with all new caps.

Of 30+ wraps, (never bought the staros version). Most being deployed before the war1 release, have only had lightning failures and dead CF. Still got a few at over 5 years running in the same neighborhoods and surge protection, power grids as war1s.

I want lifetime licenses on quality cpe this time around. Problem with war1 is it was the only easily available option as software upgrade paths. I don't mind mounting boards if i have to. But how can i ever trust compex again.

[DrLove73]

Maybe in the future there'll be an option for us to compile and update binaries on our own and that would take some of the workload off them, as it concerns fixing things like the DHCP daemon. That way those of us who have the ability can quickly fix things like that, and the number of needed releases and compilations across all the platforms would be smaller.

Even just a tool to update binaries in image files, binaries we download and install ( must have a security hash file to integrate or something? ) into our images and then we update our systems... More work for us, but maybe less need for them to update versions.
No, that is not going to happen. they are protecting their driver by encrypting the file system.

The reason for not updating 1.3.x is that they wanted to move from 1.3.x branch. And also they have very low upload in Thailand on ADSL they have there. So uploading all versions takes a lot of time. Not to mention compiling time. So they decided to halt 1.3.x development and force 1.4.x. If there was no problems with 1.4.x driver stability, we would happily enjoy latest build. And we still may.

My course would be to separate driver updates from other packages if possible and to release 1.3.24 with all non-driver improvements so there is one fully stable release. That would ease up situation a little. Also, using something like NoMachine GUI access via ssh to Valemount's "remote server/workstation" with plenty of CPU power and upload bandwidth to be used as compiling server could provide much easier and faster compilations.

[Kevin Gerrard]

west texas here and we bring the heat, the hotter n hell 100 bike race is here in the middle of august, we dont see the heat as of now.

not at all, 300 cpe's and 35 of the war's, metro's, quad and dual gateworks, a couple of wraps

and even an old two port war from somewhere weve had for ever acting as an ap

never noticed the heat yet

hope it dont start, lol not in this area anyways
whew

[Kevin Gerrard]

we have lost 90 percent of the boards weve lost because of either water running into the poe (hole in cable jacket) or they dont have it on a battery

if it blows up twice then we ground that sucker other wise we dont even do that to the cpe's

and it usually dont rain here making the ground ripe for the electricity and static ( 18 to 23 inches a year)

course we are breakin records right now
lol

just some input

[Stratolinks]

Companies that listen to their customers and meet their needs continue to have customers.

Yes you are right. Companies that listen to small groups of customers will continue to have small groups of customers. Companies that listen to the needs of the larger base will have a larger group of customers.

Personally I am very glad that Star-OS is looking to the larger base of customers. A large customer base is definitely preferred over a small customer base.

[Mark]

If you do not deliver DHCP service to customer routers (esp Netgear), then you will not see the bug.

I deliver DHCP to all my customers. Well, perhaps not on a couple, but that's about it. So, there's a LOT of them out there, and I've never seen any fatal bugs. Other than auto-auth taking absolutely forever to assign an IP, I have observed nothing. My access points all serve dhcp to the client radios, using static assigned IP's (per mac addy).

Wonder what I missed... :confused::confused:

[Mark]

Yes you are right. Companies that listen to small groups of customers will continue to have small groups of customers. Companies that listen to the needs of the larger base will have a larger group of customers.

Personally I am very glad that Star-OS is looking to the larger base of customers. A large customer base is definitely preferred over a small customer base.

Over the last two years, both participation and feedback here has declined tremendously. It used to be that hundreds of messages were posted in a busy day. Now, I've seen days go by with only a few, enough to count on one hand.

I'd have to take exception to your statement, though. It doesn't work as you'd like to imply it works. Businesses that listen to their customers...and apply what they learn with wisdom do well. Businesses that just respond to the loudest voices and do so without discrimination... fail rapidly. Business is not about democracy, it is about leadership. Failure to lead = failure in business.

[handyman]

I deliver DHCP to all my customers. Well, perhaps not on a couple, but that's about it. So, there's a LOT of them out there, and I've never seen any fatal bugs. Other than auto-auth taking absolutely forever to assign an IP, I have observed nothing. My access points all serve dhcp to the client radios, using static assigned IP's (per mac addy).

Wonder what I missed... :confused::confused:

Auto-auth? I don't believe the auto-auth feature is involved.

Do you use isc-dhcpd out of the system->services menu, for statically assigned addresses? That's where the problem is. The failure depends on how the client runs the protocol. Client radios may be immune, but a nontrivial minority of Netgear routers trigger the bug. I estimate it affects around 10% of the Netgear routers on my network. The problem is well-documented on these forums and at ISC, and is completely reproducible for an affected router.

[go.fast]

Doesn't feel like we are listened to.

[Stratolinks]

Over the last two years, both participation and feedback here has declined tremendously. It used to be that hundreds of messages were posted in a busy day. Now, I've seen days go by with only a few, enough to count on one hand.

There are many reasons for the activity on a support forum to dwindle. The primary reason is there is a stable product that has had any issues reported (yes, not all have been fixed) and there isn't much left to say about it. The busiest times are when there is a new release and everyone is finding the little bugs, the giant bugs, and the nuances with the new release. Most of the discussion in the last year or so is about other things, not about the latest release's issues.

I'd have to take exception to your statement, though. It doesn't work as you'd like to imply it works. Businesses that listen to their customers...and apply what they learn with wisdom do well. Businesses that just respond to the loudest voices and do so without discrimination... fail rapidly. Business is not about democracy, it is about leadership. Failure to lead = failure in business.

Not far off what I said anyway. Lonnie and crew have listened to many feature requests and had open discussion about what we need.

Personally I trust that they will bring out a balanced set of features that will fill the needs of the majority of us. Not everyone will be satisfied, but once this next major release is out the door, I am sure there will be much discussion on the forums then. I also expect that further updates will progress at a decent pace again after the monumental task of the next major OS update is done.

[Kevin Gerrard]

first it was a teletronics cpe, couldnt afford those so we went to cards in pc's and lmr 400 coax, then it was the big mistake tranzeo, then the cb3's (try puttin a bunch of those on an antenna, wont get more than 30 if there is any users at all with those), finally our customers are happy, but yes we did take a round about way of getting here, and yes they put up with our learning curve and ignorance.

do not think that valemount dont see things changing faster than alot of us, trust them to the future, past results have been stellar here for us

[Mark]

first it was a teletronics cpe, couldnt afford those so we went to cards in pc's and lmr 400 coax, then it was the big mistake tranzeo, then the cb3's (try puttin a bunch of those on an antenna, wont get more than 30 if there is any users at all with those), finally our customers are happy, but yes we did take a round about way of getting here, and yes they put up with our learning curve and ignorance.

do not think that valemount dont see things changing faster than alot of us, trust them to the future, past results have been stellar here for us

My experience is that Valemount's moving slower, and that the environment is changing faster. I know they're not really Valemount anymore, but hey, it's a cool name.

I started my ISP shortly after the introduction of atheros support, and I've been all Star-OS, save a few other items that have wandered by from time to time, until around the time the WAR boards were announced, and before the WAR-1 became widely available. During that time when there was a dearth of equipment, we tried many things and it was easily recognizeable that nothing could replace it at the time.

When I no longer had a reliable vendor for WAR-1's that wasn't either extremely slow to get to me, or cost a fortune in shipping, or simply wasn't reliably in having equipment in stock, I tried other stuff.

WAR-1's @ 5 ghz have nothing whatsoever on certain other equipment that's a convenient sealed device, in terms of performance. But ther "other" was so incredibly less costly, and because of that, i've had my best growth, even in a recession economy.

the point of all my posting, was that when this became Star-OS, the technology that supported the commodity atheros chipset was non-existent, and we blazed that trail. Today, the Atheros cihpsets and support for them, is now commodity. Variable in quality, but still commodity.

What is now the largest factor, is having a business model that competes in a very different world. We were once the only possible choice in a huge area. Then we became the best priced choice in a huge area. Now we're just a choice in a huge area, with small regions of exclusivity. OUR world has trasformed around us. For me to say that my old business model is still competitive... is to lie to myself.

I'm sorry, but my original "using inexpensive wireless equipment to spread broadband" model is now commodity as well. People with a small fraction of the experience and knowledge I had to gain the hard way are now rushing in to compete, becuase others have conquered many of the challenges for them and offer them in neat, pre-packaged bundles.

It's evolution time for me. Again. I intend to NOT become obsolete... Nor my means of competing the free market.

This is how and why I started this thread, as I look at where things have gone. The market space for a $300 cpe based seller of commodity bandwidth... Is crowded to the hilt. Business models have been devised to render that price irrelevant to the large providers, and force small providers to find other ways to compete.

WE ARE ALL, save perhaps 10 or 20 total users of Star-OS, small business, without deep pockets investors, and almost all of us now face at least one form of competition and we're going to face more. Do we now relegate ourselves to the ever small niche markets where we can get a consumer to give us $300 up front? Or do we change?

You ask me to "have faith". I have faith that Lonnie, Tony and crew will write great drivers eventually for commodity wireless chipsets. But that effort is one that by fiat, restricts the business model(s) to the ones we used 5 years ago. I'm asking for evidence that they understand how technology can and does constrain business...and that the business they are aiming at is, by the nature of its own success, evolving into non-existence.

Lonnie has not addressed ANY change that suggests any different approach to the business model than that which was devised almost 10 years ago, by them and other pioneers in this business - and adopted by me 6 years ago. This is what I have asked for, and instead, the answers have been more along the lines of confirming technological constraints on business model, rather than looking to achieve a higher level of flexibility.

It involves THEIR evolution from being premier code-writers and software integrators... into either software only producers to install on other's commodity equipment... or produce commodity equipment competitive with other's efforts to support the competitive business model.

I can say without hesitation, that 20 or30 dollars around the edge of the price of 'stuff' isn't the issue. There isn't any magic price, under which we become magically successful. No, there is just an area where rational models work, and a area where it does not, with a smallish gray area in between.

Originally, thier model was to build software, sell it, and the users installed it on whatever hardware it could be made to work on. Then, they switched to selling the hardware and software bundled, but the essential model being unchanged. The change was because they attempted to sell that commodity hardware, and their model wasn't efficient enough to be price competitive with other vendors. Lonnie complained loudly and vehemently about how we were supposed to just give him more money, for the exact same part, out of loyalty.

Just how much were we supposed to pay Valemount out of loyalty, to line their pockets (removed from our own) in order to satisfy his need for us to be loyal? That question was never answered, but we apparently failed, none the less.

So, began the effort to irrevocably tie the software and hardware, so that "loyalty" was never the question, and that Valemount would always be able get the profit from both software and hardware - no "loyalty tax" required.

This resulted in massive pains for many of us, due to teething pains, and the inevitable pains of change and growth. It's just how life is. And it lead to many angry and fearful people when supply could not meet demand, and many of us were left in the lurch, hurt and frustrated, when the people who kept saying "trust us" hurt our business as we tried to maintain loyalty.

There are many kinds of trust. I do trust they conquer the software demons. And, that they are diligent to find and squash faults and to continually seek to improve beyond "just good enough".

And, that's as much as I'm prepared to give. Trust my future business model to thier dictates? That's the open question of the day.

All those other brands of fully integrated equipment SELL A BUSINESS MODEL - many up front. Some are implied, but most are overt in thier advocacy. Some just try to appeal to what they think is ours.

I asked the technical questions, because those answers dictate the business model, if things continue as they have been. If the physical form of products continues unchanged, so does the financial model. And it's approaching obsolete. Lonnie has complained that 'every time I make it cheaper you just want more".

In a way, that's true. We have had to stay ahead of something, all of our WISP lives. We still do. Closing my eyes and ignoring the steamroller won't stop me from being squashed.

So, one last time I ask. I have not flamed this forum, nor anyone here. I have tried to be direct, clear, and to a degree, pointed and direct.

Where is this train headed? Do you have a plan to address how WE need to compete? Or not? It could hardly be any trade secret. Your strength is your capabilities with code, not Compex (or any other maker's) making boards, nor is it your massive financial backing that makes you bust the price barriers on hardware - we know that's not there either.

So, are you going to take leadership in addressing business models, not JUST driver writing skills, or is this train headed down the track marked "you only get off here, nobody gets on" till it's empty?

I cannot be clearer, nor can I devote any more time to this. If this can't be answered, it can't. The implications of 'cant' should be stated directly, not just left. Do you need ideas? Do you need sparks to be infused? Then get people to provide them. Do you need more creativity? Then find it among us.

So, please, address this with more than a contemptous "trust us".

You haven't earned THAT much trust. Nobody in their right mind would extend that much - to anyone. We all have to answer... to ourselves, our families, our employess, our partners, our investors, our time, even. Just saying "I trusted them" isn't what independent and determined people do to succeed, especially if they failed to ask about the elephant in the living room.

Peace. I rest my case.

[lonnie]

Mark, we have consistently been telling people what they need to compete and where they want to head in order to be there for the future.,

We have not had a $300 CPE forever, so I do not know why you keep blasting me with that ridiculous number.

If I tried to dictate your business model you would not accept it, so why are you asking me to do that?

We do not need any more spark, etc, we just need time to work on the driver. As we got out of x86 land things took longer for the simple reason every piece of hardware is unique. There is no BIOS to handle things and we do it all on our own.

You are obviously unable to wait for the new, and very near solution, so I bid you farewell, and wish you all the best in your quest.

[mimbach]

WOW!
Never before have I seen a vendor tell a customer to piss off who probly spent an hour writing that post begging to remain a customer.

I come to the forum everyday with fingers crossed. I guess I'll continue to wait and put other gear up while I do.

Lonnie and Tony PLEASE I am begging you; show us all what we have been waiting for, new N gear that works.

[lonnie]

We are very close to having something in stock and ready to. If that is not good enough, then I am truly sorry, and will bid everyone who cannot wait a friendly farewell. What else can I do? It simply is not ready to ship so I CANNOT show anything, can I.

When I repeatedly say we are nearly ready but unable to ship, and the customer repeatedly threatens to go elsewhere unless we show something right now, I am in a no win situation, so the best I can do is smile and say enjoy your search. I will not lie, and being silent is not a good idea.

I have spent countless hours on this Forums advising people what their business model should be like. Some take my advice and some do not.

We have always kept pace with technology, and sometimes we have advanced it. While it is true we are not the first to market with 11n, we will be one of the best. Right now that is all I can say or do, and for Mark that was not good enough. I am sorry, but there is nothing more I can do.

We write driver code and we sell wireless routers and I advise people how to install. I help troubleshoot and I sometimes help to design layouts and choose repeater locations. We are good at what we do. You know that, Mark knows that. So why this sudden panic that we ship and show or else he goes elsewhere? Our past should be a good view to the future.

One thing we do not do is try to hit a moving target, which is what this "business model" stuff is. And don't tell me that price is not an object and then make posts to say we are overpriced (using a dreamed up $300 price for CPE). We do drivers and we integrate into small computers and get the most we can out of them.

[Mark]

Mark, we have consistently been telling people what they need to compete and where they want to head in order to be there for the future.,

We have not had a $300 CPE forever, so I do not know why you keep blasting me with that ridiculous number.

If I tried to dictate your business model you would not accept it, so why are you asking me to do that?

We do not need any more spark, etc, we just need time to work on the driver. As we got out of x86 land things took longer for the simple reason every piece of hardware is unique. There is no BIOS to handle things and we do it all on our own.

You are obviously unable to wait for the new, and very near solution, so I bid you farewell, and wish you all the best in your quest.

I take you at your word, and then you accuse me of making it up. You said that you're going to continue the "radio module in SBC" physical format, and that things would cost MORE, not less, than they have been. The number comes directly from the retail price of products. MIMO antenna, enclosure, and guesses about the board and radio prices and it comes out to $300 CPE, because there are NO reasonably priced MIMO antennas anywhere on the open market. The closest thing is to do a 1x1 (which is really just N) using standard stuff we've done before. That's no big leap, and it's MORE ( per your own word ) than what we pay now.

I never once demanded you show me ANYTHING, except to say you're going to address the business model aspect, which you still will not.

You've said you're going to N, MIMO, and then TDMA. All of these are just software. Ergo, we're stuck with the market on MIMO antennas and the rather high cost of discrete components. Lowest cost dual pol antenna I could find was 80 bucks. Lowest cost one with decent gain was 160. One radio module, one board, two pigtails + two short RF cables, or two short custom pigtails, none of this matters. the price goes OVER 300 for a MIMO CPE that reaches more than just a short distance at 5 ghz.

I asked whether you were going to address the physical form with an eye toward making slick unified products, and you said NO.

I have asked every question I can, I have every fact at my disposal that I can find is relevant. And now, you accuse me of just making up crap.

Well, if I'm wrong, because of what I just said above, then dangit, SPILL THE BEANS about the stuff nobody else but you knows, or at least stop publicly accusing me of making up what I've NOT made up.

[lonnie]

Mark, we make software which we install on units that others make for us. Slick looking does not make the best working, and you know that. In time we will improve the looks, but don't you really want it to work first?

I am not going to get involved with the business model. We'll build the router and you put it into your business model.

I am not going to talk about pricing or other features. This is a developing situation and others are having trouble making it work. Wouldn't it be nice for them if I showed them our roadmap before we had a chance to go to market. So drop this demand. You're wasting your time.

In general MIMO costs more. You need two antennas in the same physical space as one, so the gain is less and that means less distance. The radios are dual front ends, so it makes for more power amplifiers, which cannot be as big as ABG units. This gives less distance. Given these two things, yes, it will be costly go a long distance, but now, maybe you see why I have been proposing that guys go to Microcells.

You have to plan for the higher downloads people will be making. Gone are the days of 50-70 people on an AP. People now use the Internet to watch TV, and that requires HUGE bandwidth and all the time availability.

In order to get the most out of MIMO you have to work within its limits. You want a distributed Microcell and routed network. That guy connecting at 25 miles is just going to have to move closer, expect less bandwidth, or change providers. He'll hurt you if you connect him to your AP with closein people. His will be a special p2p short using high gain antennas. You'll have to charge a high fee, but then, this is the real world, and he cannot expect a free lunch. He has to pay what it costs, and if he refuses, then you simply bid him farewell but be sure to wish him good luck. If someone else can do it, then they have solved your problem.

[deano]

This discussion is starting to get irritating. :mad:

IMHO, the 1.4.x software needs to be finished, and when I say finished, I mean a stable, production release, with ALL the features of the 1.3.X version.
Please, finish what was started.

"N" hardware at this point in time, for me, is not required, and I seriously doubt that I will even need it for at least another 1+ years, and only for backhauls, based upon what I have read here, and elsewhere.

And when Lonnie says microcells are the answer, he is right. It works great.

[tkerns]

Lonnie edit in red
"Wouldn't it be nice for them if I showed them our roadmap before we had a chance to go to market."

Show who? Lonnie, maybe you should take a good look at your competition... you are the LAST one to market. Maybe last to market (as I already acknowledged, but they are not doing so hot and we hope to do better. I will not discuss our approach or what we are fixing or using. No pint helping them fix theirs until we get there to show ours off.

No one is waiting to see what you are doing... they are already producing it. Over the past year you have fallen down and badly let all of us down. Reports are they are really not doing so well. It seems not everyone has been let down.

Poor quality radio's, boards, buggy software that is not being fixed... maybe you should think of moving back to Canada and get your mind back on business. or you will be gone in the wireless world. Yet some people report stellar results. My mind is firmly on business, thanks, and my health has improved tremendously, meaning my life will be longer and thus more time to focus on business. But thanks for the suggestion

I haven't purchased new StarOS hardware in over 6 months, why? I got tired of having to go back and replace a power supply or radio that came with them. I'm still installing and growing my foot print but it has been with other products, because it has what I need, and it works.

Lonnie, those on here are trying to give YOU a wake-up call.... listen to them. You have fallen way behind. Not so far behind.

Tim

[Kevin Gerrard]

Sorry but i cant help it, we are running 1.4.22r on all towers, 21 towers 34 access points and all in one county, mostly 2 to 3 mile links

flawless and pulling 10 mbs up or down to 80 percent of our clients, and we even have about 25 addresses with true routables doing the vds tunnel thingy.

the only problems we have seen in a long time is always and i say it again ALWAYS with a long link, like a 6 or 8 mile link

that is why our strategy, game plan or what ever you call it is to put em on and when we get 10 or 15 24 db antenna customers we put a tower in that area and boom short range antena and boom that ap is now 8 to 10 mbs

yes we have issues but mostly my own ignorance,

the whole industry is changing we just put a verizon fiber in at a place in dallas and it was 30 mb down and 15 up for 250.00

tough to beat that eh? but we dont have to and it aint the software that is hurting us there it is the fact that we cant purchase the bandwidth that cheap and here in wichita falls, texas i would challenge we can keep up with and prolly even beat em all and that includes time warner cable

and yes we have a backhaul that is pulling heavy right now so we took the towers behind that link and routed a few of them through a different route, nice to have so many towers in one area, we are looking for total dominance in our county since even 1 percent of 200,000 monthly is a large number

lol

end of memo and now im in trouble i guess i revealed our game plan
shoot

winks

just my 2 cents worth

yes peeps that dont know what they are doing want a gui
then use windows and see what it gets ya

[Kevin Gerrard]

been awhile since i churned some butter, winks

[go.fast]

Are you using XR9's?


Sorry but i cant help it, we are running 1.4.22r on all towers, 21 towers 34 access points and all in one county, mostly 2 to 3 mile links

flawless and pulling 10 mbs up or down to 80 percent of our clients, and we even have about 25 addresses with true routables doing the vds tunnel thingy.

[Mark]

Lonnie edit in RED

Mark, we make software which we install on units that others make for us. Slick looking does not make the best working, and you know that. In time we will improve the looks, but don't you really want it to work first?

that's why I am not asking and have not asked you for an ETA. That's why I have not said a single critical word about how long it takes or doesn't take to get done. I am simply pointing out that this is a technological change, and with that change, combined with the need to reinvent some wheels on your part, is an opportunity to address improving the business model of your customers.

I am not going to get involved with the business model. We'll build the router and you put it into your business model.

But it is an obsolete business model. Instead of lowering the cost of clients, you're talking about raising them. That's been my point all along. You kept saying this was not the point but always ended up saying it anyway. Good to see you make a stand. We have not released details, especially about pricing, so perhaps you are being too quick to slam us with high priced clients.

I am not going to talk about pricing or other features. This is a developing situation and others are having trouble making it work. Wouldn't it be nice for them if I showed them our roadmap before we had a chance to go to market. So drop this demand. You're wasting your time.

I don't think there's ANY other vendor which hasn't already set thier models and maps on paper. The real expensive ones lay it out in great detail. No fear someone else will steal them. I wonder why. Maybe it's becuase they're not worth used toilet paper to competition. Brand M, Brand U, and others are NOT waiting to see what you do before they move on. They're already in a headlong rush with their own ideas. Some are to market already. We're just waiting around. I stopped advertising. I've stopped deploying new sites, I'm making bad excuses to my customers, waiting to see if I can continue down the road I'm on, or if I have to change. I know many are not waiting to see what we will do, and we hear they are maybe rushing new tech to market that is not so good. I suspect you'll see them make some changes after we come out, which is why I do not want to give anybody a game plan so they can change and adapt before we get a chance to show off a bit.

In general MIMO costs more. You need two antennas in the same physical space as one, so the gain is less and that means less distance. The radios are dual front ends, so it makes for more power amplifiers, which cannot be as big as ABG units. This gives less distance. Given these two things, yes, it will be costly go a long distance, but now, maybe you see why I have been proposing that guys go to Microcells.

Which is all fine and good, but going to microcells does several things.

1. Removes the need for high bandwidth AP/client systems. We can easily feed 20 meg to 8 to 10 clients in a microcell, using standard 11a, and el cheapo clients from brand x, for 70 bucks. And we won't outgrow that for many, many years. So you see that is good.

2. Raises the ratio of infrastructure vs clients. What do you think it REALLY costs to build a rural access point? this makes affordable clients even MORE important.

3. What do you do when city/state/county laws basically prohibit such deployments, by restricting deployment to specific already existing towers/buildings/etc? Break the law? Get fined? Go out of business? You have to fight those battles. It is a lot easier to rent than own.

You have to plan for the higher downloads people will be making. Gone are the days of 50-70 people on an AP. People now use the Internet to watch TV, and that requires HUGE bandwidth and all the time availability.

Actually, the point of having an ap and client that can do 100 mbit is so that we CAN have 40 to 60 clients on an AP and still deliver their constantly rising demand for bandwidth. Sure, we need backhauls and infrastructure, but those exist in large numbers already, from a wide array of sources and a wide array of prices. 11a backhauls are no longer enough for a busy 3 sector AP. Moving to 11N backauls would simplify deployment and reduce that backhaul cost a little, but that's about all it does. Either we need a 3 - 5 X quantum leap in the ap-> client throughtput, or we expect to move to a microcell...where it can be done, but that move, with more ap's, has to be offset by using very inexpensive CPE so that our investment / customer doesn't spiral out of control and we just become dried bones and skin, pecked to death by competition. You miss the point that soon almost every client will be doing 24/7 high bandwidth traffic, and that simply rules out a large number of clients, no matter how high the AP bandwidth is. I've said this before and you ignore it, but at the same time blast me for not addressing my clients business model. You only make changes when it suits you or hits you on the head. But you always blast me for causing your trouble. I try and be patient, but my advice is typically ignored and you argue with me to try and prove you are right. I always ask myself --> if Mark is right, why is he having trouble?


In order to get the most out of MIMO you have to work within its limits. You want a distributed Microcell and routed network. That guy connecting at 25 miles is just going to have to move closer, expect less bandwidth, or change providers.

We're supposed to alter the time/space constants? shrink distances magically? Doesn't work in non-dense populations. My suggestions are sound. You cannot ignore real life constraints.


He'll hurt you if you connect him to your AP with closein people. His will be a special p2p short using high gain antennas. You'll have to charge a high fee, but then, this is the real world, and he cannot expect a free lunch. He has to pay what it costs, and if he refuses, then you simply bid him farewell but be sure to wish him good luck. If someone else can do it, then they have solved your problem.

No, they haven't solved anything, they've taken my customer, because they're not constrained by an obsolete business model. This is NOT how you compete. This is how you plan to go broke. Again, you miss the point. They cannot provide high bandwidth from a distance either, so either they are closer or they talked him into settling for less. If you cannot serve him without hurting your other subscribers, yes, the problem is solved if he moves to your competition.

[Ick]

I'm with Deano, this discussion is getting irritating even for me.

All we're looking for is a 1.4X stable working release, then worry about N.

[Kevin Gerrard]

no gofast we use the cards that ship with the boards from star-os

the 10 mb's would be total for the ap so if many are using it of course then it doesnt but for the most part its golden, we shape business accounts to 5 megs up and 5 megs down, the residential get 2 megs up and 1 meg down.

even the residents never complain

[Kevin Gerrard]

even got a 50 foot tower at the lake with 44 peeps on it right now, 2 mbs down and 1 up to all of em, of course never see more than 7 or 8 using it

the new tower out there is 160 foot and sectionals, waiting on electricity

damn rain

[SpecialK]

Mark,

We are all patiently waiting.
I too have been putting off upgrading/expanding customer base because I feel the current product is not good enough. We have too many bugs/problems with the current firmware.
But to really stick it to Lonnie is unfair.
Yes it would be nice if he would share more info. but he feels somehow this will give the competition an unfair leg up. This is his choice.
I also disagree with that theory.
In my mind the real competition to StarOS is already deploying N based products. And it only makes us all look to try the alternatives.
I watch several other forums/lists and I see almost everyone on this forum trying other stuff including you.

If you really need stuff use your alternatives.
You clearly have been trying them as well.
So if it was working so good you would go silent and use the other brand and maybe check back when Lonnie has progressed his products.
The other stuff is so cheap I would rather get a customer and replace the gear later if need be than lose them. You won't get that chance again usually.
On another note where do you think you are going to get 100 mbits from an AP? Maybe with a handful of customers but even if the new N MIMO devices can do a theoretical 100 mbits it will be using 40 mhz of channel space. So you can have a whopping 2 non-overlapping channels. What about competitors? What about your own other towers/sites that are close enough to hear these? Sure a PtP link will end up doing 100 mbits with 2' dishes or bigger to help keep noise down. But you are living a pipe dream if you think that huge PtMP cells with customers 10 or 20 miles out are going to work.

That 100 mbits is a pipe dream in 5.8 Ghz unless you are so in the sticks.
But even there all it takes is one rookie to take a few rockets to go across the street and you are all done.

The cheap $50 - $100 radio will very shortly be the double edge sword for us. As every computer geek on the planet will either make a mess of the airwaves for us or be our competition. It makes wireless as a whole a commodity offering.

[skyclimber]

We have actually close to 2500 StarOs customers. We did try many brand U and M gear. From the numbers we've seen, the 802.11N stuff is not ready anywhere. TDD on 802.11a would be a better option for long shot backhauls based on 2 differents products we got from L brand.

Wimax and mini-pci 802.11N radios will move to a next generation soon and all customers for brand U and M will have to be changed for new one. Waiting is not so bad if better is coming.

[go.fast]

no gofast we use the cards that ship with the boards from star-os



The XR9's are 900MHz. You don't use star and 900MHz?

[Kevin Gerrard]

nope, no 900 here, the city police do something with the cars in that area of the spectrum and it doesnt work right either, thank god it aint my stuff, lol

we took a 12 db to 12 db yagi horizontal out to my partners house and laid one antenna on his truck flat and i took of down the dirt road, went 6 miles up and down hills and it pinged the whole time,

same senario on two good towers and it wasnt worth a crap, got me but being just two of us in this race so we didnt waste much time with it either

[Mark]

Mark, Lonnie edit in RED

I watch several other forums/lists and I see almost everyone on this forum trying other stuff including you.

I'm still hoping for a decent distributor for STar-OS to come along. I'm basically restricting my off brand stuff to small deployments, where the performance doesn't really matter.

If you really need stuff use your alternatives.
You clearly have been trying them as well.
So if it was working so good you would go silent and use the other brand and maybe check back when Lonnie has progressed his products.

it isn't that easy. It is easy if the other works to your satisfaction. If their stuff is not what you want then stop saying we are behind because we refuse to release not-ready code.

And, the problem is, it DOES work. The difference in performance between Star-OS stuff on the cpe end and some of the other brand stuff is... nil. Then why do you limit the off brand stuff to small deployments, where the performance doesn't really matter. I just put in a brand U backhaul and in a 20 mhz space, it will move more than 65 mbit. That's just all the traffic I could figure out how to generate, and yet we apparently had headroom left. That's 3 times what a turbo A star-os based link would do, over the same distance, same space, from same mast to same mast. And it doesn't rely on "compression" to achieve the speed, either.

I want to move forward, with a sensible and economical plan, and I'd prefer to stay with what I got. But after being told that X and y do not work, when they clearly do, I lose faith in the people who haven't been straight up with me. And then, they say "trust me". Maybe the newer stuff works, but clearly people have had trouble in the past. You cannot deny that, and you cannot refute what I have said because someone improved their units, which is what should have happened years ago, rather than people busting our balls to make the crap work with our stuff

And, then after saying "costs will go up", publicly excoriate me for taking it as fact, saying to me in effect: "it's not out yet, how can you say that?" Dangit, GET YOUR STORY STRAIGHT. Lonnie has contradicted himself in this thread alone, on that topic. You put a specific, very high number to the unit. I am saying we have not released pricing, so quit blasting us for a $300 CPE, which is what you would be able to put together.


The other stuff is so cheap I would rather get a customer and replace the gear later if need be than lose them. You won't get that chance again usually.

No, you're right. But then, if I change direction, it's probably as permanent as this one has been - nearly 6 years.


On another note where do you think you are going to get 100 mbits from an AP?

Because I think you can, and I think you can do it in a 20 mhz dual pol channel. If you can do 65 with hobbled 802.11, then 100 or something reasonably close, using TDMA or TDD or some other efficient method, using radio rates of 130... Seems perfectly within reach to me. The radio rate is there, the overhead need not be 50%, like it is with Brand U's "M" stuff. Once again you seem to indicate they are clearly not ready and certainly not superior. Quit busting us for waiting to release good code.

Maybe with a handful of customers but even if the new N MIMO devices can do a theoretical 100 mbits it will be using 40 mhz of channel space. So you can have a whopping 2 non-overlapping channels. What about competitors?

In much of my space, I don't have any. I do where there's lots of people, though. So, robust non-802.11 protocols have to eventually come into play, along with completely viably priced whitespace and 3.65 stuff.


What about your own other towers/sites that are close enough to hear these? Sure a PtP link will end up doing 100 mbits with 2' dishes or bigger to help keep noise down. But you are living a pipe dream if you think that huge PtMP cells with customers 10 or 20 miles out are going to work.

It has to work. there is no alternative. If it means I turn everyday operations over to my partner and spend a year hacking my own driver, then I'll do it. There is no such thing as 'can't be done', only "I don't want to do that". Why don't you hack your own driver then? It surely must be easy. You must still be a teenager, you think you know everything.

That 100 mbits is a pipe dream in 5.8 Ghz unless you are so in the sticks.
But even there all it takes is one rookie to take a few rockets to go across the street and you are all done.

That risk is born with ANY model, except for a fully robust system with re-written protocols and tight engineering.

The cheap $50 - $100 radio will very shortly be the double edge sword for us. As every computer geek on the planet will either make a mess of the airwaves for us or be our competition. It makes wireless as a whole a commodity offering.

Did you not think that day would come?

It's not just come, it's already here, in play and we're needing to beat it, financially, performance, reliability, and in terms of being good business and PR people.

We need to at least get ON PAR, if not ahead, sometime in the near future.

[go.fast]

Ok, so thats why we can't use 1.4.22r, XR9's work with 1.3.23 but not 1.4.22r

so for us that use 900MHz, 1.4.22r is out and we don't think it was finished.

And for the record, like you, I only buy Star gear and only use the cards that come with the units, except for XR9's.



nope, no 900 here, the city police do something with the cars in that area of the spectrum and it doesnt work right either, thank god it aint my stuff, lol

we took a 12 db to 12 db yagi horizontal out to my partners house and laid one antenna on his truck flat and i took of down the dirt road, went 6 miles up and down hills and it pinged the whole time,

[DrLove73]

Any ideas? My business partner is ready to lose his mind and force dumping StarOS, as we've been doing this business for over 5 years, and until this last year, we've seen nothing like this at all. I've found and fixed leaky connections, leaky antennas, bad cables, bad cars, bad power supplies, bad cpu boards, interference, self interference, and many other issues.
Mark, when do you think your partner is going to be satisfied with you pressuring Lonnie? My belief is that your problem is lack of proper design and recommended practices rather then business model issues. I bet that if you were able to fix problems on your network, both you and your partner would be much more relaxed and actually happy with StarOS (except for known bugs/problems that need further work).

As for further development, cheap units are not the feature, at least not on the long run.
For higher speed you need more powerful hardware. More power warrants bigger price. 11n and newer units will be more expensive. WAR1 was a response for cheaper CPE and it can only push some 5Mbps? because of weaker CPU). But still people try to use it for AP's and similar applications. You will never have it both ways.

Take a look at these screenshots:
http://www.plcomputers.net/download/Speed_to_2Km_away_router_on_5GHz-1.jpg
http://www.plcomputers.net/download/Speed_to_2Km_away_router_on_5GHz-2.jpg
http://www.plcomputers.net/download/Speed_to_2Km_away_router_on_5GHz-4.jpg
All 3 units presented involved in this test have connection tracking on, 2 of them have packet marking and policy routing active and 2 are used for PtP VDS link to my upstream provider.
First two screens are Rx and Tx (starutil built-in test!) to site 1,5 miles out on a busy silo (4 wireless providers with ~10 5GHz antennas) and mine there is small 21dB grid. It was on rate 54 and -62 signal.
Third is Rx and Tx simultaneously to two X86-PC boards there (only one test can be performed at one time on one unit). It was on Turbo rate 108 and -65 signal (if I had not tried this tests, I would have left it on rate 54). Note the CPU on middle right unit is 100% on following CPU:
---[ CPU Information ]--------------------------------------
processor: 0; vendor_id: GenuineIntel; cpu family: 6; model: 6; model name: Celeron (Mendocino); stepping: 5; cpu MHz: 501.139;cache size: 128 KB; fdiv_bug: no;
hlt_bug: no; f00f_bug: no; coma_bug: no; fpu: yes; fpu_exception: yes; cpuid level: 2
wp: yes; flags: fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr; bogomips: 1003.04; clflush size: 32
50% of that CPU is used to pass traffic to other unit. Other unit is connected to my office via LAN, so those 10.000 KB/s might be higher if CPU was strong enough. (I do not need such traffic except to allow my brother to download files from office PC with high speed. And HDD speed is issue there. SATA HDD 320GB max out around 250MBps of traffic).

My own network (6 omni's+links) uses ~3Mbps and they are still on 11b mode with 100% Alien units. Business climate and my resources are not for any expansion so I lay dormant with wireless and wait for a lot of things to clear out. But those alien units are not the cheapest ones I can find, nor they are PCI radios with long cable as 80% of providers here install just to attract new clients. I buy ones with built in NAT and with fairly good PSU, so only ~10% of all customers have any problems during a full year, maybe less. that is why I have 90% of large companies around here on my network and they are all happy with me even compared to ADSL and cable operator as competition. "It works". So I provide cheaper units but I am still the most expensive option here non the less. This is magic circle.

6-7 months ago I arranged for transfer of small town ISP (their hardware, my brain and upstream access) to migrate to another network (My old upstream ISP I ditched about the same time). They were hogging my upstream bandwidth with p2p and game playing at the same time, and my profit from them was practically down to nothing. And any unsatisfied customer (cry-babies compared to the rest of my customers) would constantly bother me and whine all day long. 70-80% of problems was upstream link braking (gamers) and me throttling p2p. So I was VERY happy to see them go. I was reborn man after that. In last few months I used biggest cable operator here, in next town) for http traffic since I was not able to cope financially with their high bandwidth demands. Only month after the change I heard those whining customers started to "miss me". They realized that "gold is not everything that shines".
A month ago they asked me to "come back", to use my StarOS network to provide them access to that big cable operator (on mine account), and to again install them StarOS back on their X86-PC AP's. Their words were "StarOS is StarOS, nothing compares to StarOS." We struck a deal and they and their customers are much happier then they were with my old upstream ISP that promised them higher bandwidth packages for lesser money but failed to explain that they have progressive (automatic) bandwidth downgrade for large consumers and that not all services (ftp, p2p) have full paid for speed until it was too late.

During all this time I am an WISP (since mid 2005, with StarOS from 2006) I have NEVER installed any flaky CPE or PCI to anyone (without them really insisting on it), and any unrealistic user I encountered I even deliberately directed to my competition. But that unrealistic person is not able to say that my network is peace of s*** just because they expect too much from wireless or cheap hardware, or they do not have possibility for stable and quality connection.

When I started WISP business I was told that I need more power to provide quality service to my customers. I still have Prism PCMCIA 200mW + 1W and 2W amplifiers safely tucked away since I accepted Lonnie's advice to go with smaller power and stronger antennas and shorter cables at CPE side. And my network runs much better with CM9's then a lot of other networks here with higher power radios.

My suggestion is that you hire someone from this forums with good track records, and possibly closer to you so he can inspect it and give you pointers in what you should (generally) do differently to maximize performance AND STABILITY of the current hardware and suggest you the direction you should pursue.

Do not forget that Lonnie has committed (as I understood) to support several Routerboard models to allow cheaper and more stable (in stock availability sense) hardware for those that are willing to trade performance for lesser price and those that have MT hardware but would like to use StarOS. And if they accepted anything that I and some other members here suggested for next generation hardware, it will be much more usable and versatile platform then competition in same price range, and will actually set a trend for all other vendors. Let's wait a little more and see.