Was that Mimosa AP's or Clients and how do they compare performance wise?....................................................
I was about to cancel my 200 Mimosa client order but I am happy now I didn't.
I hope Mikrotik finds a solution before we are forced to upgrade more of our P2MP networks. Or does is show again that cheap can't do the same as expensive?
Is hoping that my present 800 mikrotik clients network could be upgraded to a fully syncd and download ratio configured network for free indeed a free ride on a pink cloud that drops you the moment you wake up?
CPE's.Was that Mimosa AP's or Clients and how do they compare performance wise?....................................................
I was about to cancel my 200 Mimosa client order but I am happy now I didn't.
I hope Mikrotik finds a solution before we are forced to upgrade more of our P2MP networks. Or does is show again that cheap can't do the same as expensive?
Is hoping that my present 800 mikrotik clients network could be upgraded to a fully syncd and download ratio configured network for free indeed a free ride on a pink cloud that drops you the moment you wake up?
You maybe right about aircraft triggering the DFS scan, why aircraft would even consider using frequencies in (or in close proximity) to a unlicensed band defies logic?And the compulsory DFS channels are a pain..... I don't know what they'd exactly pick up but the Mimosa AP swaps channels at times due the Radar detection (from where? There is no Radar to be found here!). Now this would not be such a problem if the CPE's would follow quickly. But the new channel of use will be 'radar' tested too and only after that the CPE's are allowed to 'talk' to the AP. That can take up to 15 mins. During all that the clients have no internet!
Since the DFS only works on one of the Mimosa AP's (I have 3 working and running now) I developed a theory that it are air-planes that give the problems. Since the Radar only is detected sometimes, sometimes a whole day nothing.... it cannot be a typical Weather radar beam (Rain detection radar) since that works 24/24.
But we are in the flight path of descending planes going to a nearby busy airfield.
On google I found that some plane can have installed approach/weather/ground detection radar devices that are sending a strong powerful beam straight forward and down from a plane that can work in the 5Ghz range.
This could explain the random occurrence. It would be a relative narrow beam, but with high density and energy, that only at occasions just happens to run straight over my AP....
The flight path is only used in certain wind directions at the airfield and since planes still fly from wherever they came towards the approach radio beacon and they are still at some 3000 feet above us but descending it could be that only some planes with a specific radar frequency in a specific flight path will have its radar bundle hitting our Mimosa.... lucky me..
Otherwise I can't explain the random, and at times non existing, DFS mode kick-in from the Mimosa.
I already asked them to make something like an 'override' (= illegal, but what the hack, I need to make money here..) button. But they consider themselves very law abiding and are not even willing to consider to have such an option afraid as they are they might loose CEE certifications......
But just tonight I have swift the working frequency some 150Mhz lower after re-arranging some frequencies of my other towers and see it that will improve things.
It that performance Mimosa AP to Mimosa CPE or Mimosa AP to Mikrotik and is that AC or N ?Capacity wise, when it all runs, the Mimosa is outperforming the Netmetals in both amount of connected devices as well as top speeds by at least 200%. So in that respect I am happy.....
(I can connect up to 70 clients to a Mimosa and have several units run at 30Mbps downloads and a total throughput over the Mimosa AP with some 150-180Mbps. Try that with a Netmetal!
And we are still in CSMA mode though! We still have 40% SXT-ac's to swap.)
When we have replaced all CPE's for the C5's I am going to test the Mimosa tdma with 60 clients.
And when that is done we are going to do a full CPE swap from a second Mimosa AP and make it run in full sync with the first sharing the same frequency and in a 80Mhz channel.
If that is all going to work I am a happy man!
First what we did is setup one Mimosa A5 and just transferred a mix of SXT-Lite 'n' and some 711 board radios to it. In the end we'd had 60 clients to that tower and one by one they all worked fine.It that performance Mimosa AP to Mimosa CPE or Mimosa AP to Mikrotik and is that AC or N ?
While speed is very important I think a more deciding factor longterm for retaining residential customers (which makes up 90+% of our customers) will be cost effective media content delivery, that is who can give streamed sports and movie channels, etc plus broadband in a inclusive price?....................................
I am merely going to avoid fibre coming to us.. to expensive for us as a small business and if I can outdo the present fibre competition with wireless why not!
Ok, it still has to proof itself.... but I have the feeling we are on the good track...
We are just deciding do we offer 50Mbps or straight away 100Mbps to our clients!
Well, our market is still 60% foreigners. 85% of these are British. That national content cannot be legally served by us. Our Spanish clients have little demand for streaming TV and we even get little requests for phone calls. Triple play that is big in 'cable world' is not that big in rural Spain. Most people have their mobile and can pick their standard TV by free digital terrestrial. Any provider offering triple play has a price tag at least twice as high as ours.While speed is very important I think a more deciding factor longterm for retaining residential customers (which makes up 90+% of our customers) will be cost effective media content delivery, that is who can give streamed sports and movie channels, etc plus broadband in a inclusive price?
Already we have seen this happening as sports channels offer their TV channels with broadband plus streamed content available in a bundle price and this where the real challenge lies for WISP's.
Thanks for excellent post Rudy. Sounds almost too good to be true. What is the main reason you think the mimosa can give so much more aggregate throughout. Software or hardware. How do the netmetal hardware specs compare with the mimosa ap you are usingWell, our market is still 60% foreigners. 85% of these are British. That national content cannot be legally served by us. Our Spanish clients have little demand for streaming TV and we even get little requests for phone calls. Triple play that is big in 'cable world' is not that big in rural Spain. Most people have their mobile and can pick their standard TV by free digital terrestrial. Any provider offering triple play has a price tag at least twice as high as ours.While speed is very important I think a more deciding factor longterm for retaining residential customers (which makes up 90+% of our customers) will be cost effective media content delivery, that is who can give streamed sports and movie channels, etc plus broadband in a inclusive price?
Already we have seen this happening as sports channels offer their TV channels with broadband plus streamed content available in a bundle price and this where the real challenge lies for WISP's.
We do offer the other stuff but our strategy is more in supplying high speed stable internet for the same prices as the competition does for low to medium speed.
Two nights ago I changed the first full Mimosa AP with 38 clients to its proprietary protocol and wow, that rocks... I was already impressed by its csma protocol but now with tdma signal levels increased, quality of the links increased and I hardly see any more interference.
The less interference issues is probably also to do with the somewhat higher gained CPE's and the shield they have. All levels of CPE's to the AP and vice versa are better then -50dBm so I could set the noise filter to -54dBm which means any other signal not stronger then that gets filtered out as well.
We still have to swap some 25 clients on another tower and then hope to swap that P2MP to tdma as well and then try to sync it with that first AP. It that works I can move some of my other radios away towards the 'freed' frequency of No.2 Mimosa AP to create an even lower noise around the now combined 80Mhz (yes, it works on 80Mhz wide band now!) range of use!
Clients of the Mimosa now still get only 25 or 30Mbps but we are waiting for an upgrade of our backbone and then we are going to offer 50 for standard and 100mb for high profile contracts....
The Chipset used by Mimosa is a standard chipset from Quantenna. Just a chipmaker like Atheros. Not special designed for Mimosa and found in some other wlan APs. With the B5 we've seen it is not great with interference rejection. So using wide channels you need good separation from your own gear and from other APs (and you trash the band). Not bad but not that great as their marketing.The chipset from MT is also different then what Mimosa uses. I have to look up what they actualy were but I have been reading before (independent resources and test panels) that the chipset of Mimosa is outdoing the Atheros in use but Mikrotik, Ubiquity etc. a lot. Mimosa chipset is just much better.
Mimosa had built their radio's (chipset, OS etc.) completely new from zero with only one thing in mind; Make use of the full capabilities embedded in the new ac protocol with one goal only; beat fibre in roll outs.
I agree with AF for ptp unbeatable. PTMP is not so good. I have too much invested in MT to think about changing. We are a lot better now than we were a few years ago but still need much improvement to compete with fiber on speed alone. Mikrotik have not being releasing much on NV2 for a while. I can only hope that means things are in the pipeline. On average my NV2 APs have 40 clients. The aggregate max throughout of the AP is around 30 to 40 megs. the clients are allowed 12 megs each. All works ok and not many complaints, however, this will not be any good for much longer as more and more bandwidth is required for steaming TV services. Come on MT it's a race against time now but I have faith in you.The Chipset used by Mimosa is a standard chipset from Quantenna. Just a chipmaker like Atheros. Not special designed for Mimosa and found in some other wlan APs. With the B5 we've seen it is not great with interference rejection. So using wide channels you need good separation from your own gear and from other APs (and you trash the band). Not bad but not that great as their marketing.The chipset from MT is also different then what Mimosa uses. I have to look up what they actualy were but I have been reading before (independent resources and test panels) that the chipset of Mimosa is outdoing the Atheros in use but Mikrotik, Ubiquity etc. a lot. Mimosa chipset is just much better.
Mimosa had built their radio's (chipset, OS etc.) completely new from zero with only one thing in mind; Make use of the full capabilities embedded in the new ac protocol with one goal only; beat fibre in roll outs.
Their Omni Solution (which are 4 singlepol sectors in fact) is nice designed.
Using Atheros a lot can be done. You see this with epmp and the ubiquities. What I find most promising at the moment is the combination of shielding, antenna design, RF filtering and gps sync you see with the prism station. And a second wireless card just for scanning in background (Great tool). For ptp in 5GHz yo cant beat the Airfibers at the moment and they make an announcement soon ... And guess what is synccompatible ... To bad MT never decided to put more money/people in their wireless hardware department. But this is a business decision I cant appraise.
We stay with MT for routing/MPLS. Wireless is always a mix of vendors as no one is good with every aspect/frequency.
Our biggest concern is that over head fibre on both existing telephone + electricity poles is being rolled out and is being done by the national electricity and telephone companies, they in turn can and do sell direct to private households and commercial businesses but also have to make available their network to 3rd party internet suppliers who can then sell to the same markets, for the fibre rollout suppliers its a long term win win scenario for their investment.................
So it has to be a system capable of delivering very high througput combined with relative low costs.
Off course everybody will always say fibre is better, but when its stripped from the grants given by authorities its costly.
One of the disadvantages of fibre is in the cost/deployment ratio. You need a lot of customers in a relative short range to justify and earn back the investments.
With a wireless system that is much more easy to achieve and if the speeds to the customer can be 100Mb or higher why roll out any fiber?
In my country many areas will never see fiber or when its there comes with prices we can easily undercut and thus still have a market to make money...
As always because not every network is the same, wireless equipment from a manaufacturer that uses A,B,C labelling might work great at one site but not so good at another, but what I really don't like is when you read on sales brochure "...supporting Multi-User MIMO...." but in tech specs "Multi-User MIMO **" = enabled in future software releases....same for Collocation.....this in my opinion is nothing short of false advertising.The Chipset used by Mimosa is a standard chipset from Quantenna. Just a chipmaker like Atheros. Not special designed for Mimosa and found in some other wlan APs. With the B5 we've seen it is not great with interference rejection. So using wide channels you need good separation from your own gear and from other APs (and you trash the band). Not bad but not that great as their marketing.
Their Omni Solution (which are 4 singlepol sectors in fact) is nice designed.
Using Atheros a lot can be done. You see this with epmp and the ubiquities. What I find most promising at the moment is the combination of shielding, antenna design, RF filtering and gps sync you see with the prism station. And a second wireless card just for scanning in background (Great tool). For ptp in 5GHz yo cant beat the Airfibers at the moment and they make an announcement soon ... And guess what is synccompatible ... To bad MT never decided to put more money/people in their wireless hardware department. But this is a business decision I cant appraise.
We stay with MT for routing/MPLS. Wireless is always a mix of vendors as no one is good with every aspect/frequency.
I totally agree and also don't want to switch from Mikrotik but I am somewhat convinced when I read that the same chipset using what appears to be a barebones software can give much higher wireless throughput than Mikrotik wireless can, I also am totally surprised that NV2 always uses 6Mbps for management protocol, I think Mikrotik should also offer customised software for PTP=B, PTMP =A, CPE=C ( it appears to be working for another company) and ROS software that maybe summarised as "one size fits all" is not really the best fit for everyone.I agree with AF for ptp unbeatable. PTMP is not so good. I have too much invested in MT to think about changing. We are a lot better now than we were a few years ago but still need much improvement to compete with fiber on speed alone. Mikrotik have not being releasing much on NV2 for a while. I can only hope that means things are in the pipeline. On average my NV2 APs have 40 clients. The aggregate max throughout of the AP is around 30 to 40 megs. the clients are allowed 12 megs each. All works ok and not many complaints, however, this will not be any good for much longer as more and more bandwidth is required for steaming TV services. Come on MT it's a race against time now but I have faith in you.
Sent from my ONEPLUS A3003 using Tapatalk
I never stated Mimosa had the Quantenna chipset designed for them. But they were some of the first implementing it in their product line.The Chipset used by Mimosa is a standard chipset from Quantenna. Just a chipmaker like Atheros. Not special designed for Mimosa and found in some other wlan APs. With the B5 we've seen it is not great with interference rejection. So using wide channels you need good separation from your own gear and from other APs (and you trash the band). Not bad but not that great as their marketing.The chipset from MT is also different then what Mimosa uses. I have to look up what they actualy were but I have been reading before (independent resources and test panels) that the chipset of Mimosa is outdoing the Atheros in use but Mikrotik, Ubiquity etc. a lot. Mimosa chipset is just much better.
Mimosa had built their radio's (chipset, OS etc.) completely new from zero with only one thing in mind; Make use of the full capabilities embedded in the new ac protocol with one goal only; beat fibre in roll outs.
Their Omni Solution (which are 4 singlepol sectors in fact) is nice designed.
Using Atheros a lot can be done. You see this with epmp and the ubiquities. What I find most promising at the moment is the combination of shielding, antenna design, RF filtering and gps sync you see with the prism station. And a second wireless card just for scanning in background (Great tool). For ptp in 5GHz yo cant beat the Airfibers at the moment and they make an announcement soon ... And guess what is synccompatible ... To bad MT never decided to put more money/people in their wireless hardware department. But this is a business decision I cant appraise.
We stay with MT for routing/MPLS. Wireless is always a mix of vendors as no one is good with every aspect/frequency.
I think they all do this... you have to look deeper and basically buy the product to find out what it really does......, but what I really don't like is when you read on sales brochure "...supporting Multi-User MIMO...." but in tech specs "Multi-User MIMO **" = enabled in future software releases....same for Collocation.....this in my opinion is nothing short of false advertising.
With 12 meg you are fine with MT. Problems start with 20. Customers get this speeds with udp but they complain with tcp. As speedtests are tcp ... I am so bored with this speedtester calls. This problem is for years now and consumed a lot of my time.I agree with AF for ptp unbeatable. PTMP is not so good. I have too much invested in MT to think about changing. We are a lot better now than we were a few years ago but still need much improvement to compete with fiber on speed alone. Mikrotik have not being releasing much on NV2 for a while. I can only hope that means things are in the pipeline. On average my NV2 APs have 40 clients. The aggregate max throughout of the AP is around 30 to 40 megs. the clients are allowed 12 megs each. All works ok and not many complaints, however, this will not be any good for much longer as more and more bandwidth is required for steaming TV services. Come on MT it's a race against time now but I have faith in you.The Chipset used by Mimosa is a standard chipset from Quantenna. Just a chipmaker like Atheros. Not special designed for Mimosa and found in some other wlan APs. With the B5 we've seen it is not great with interference rejection. So using wide channels you need good separation from your own gear and from other APs (and you trash the band). Not bad but not that great as their marketing.The chipset from MT is also different then what Mimosa uses. I have to look up what they actualy were but I have been reading before (independent resources and test panels) that the chipset of Mimosa is outdoing the Atheros in use but Mikrotik, Ubiquity etc. a lot. Mimosa chipset is just much better.
Mimosa had built their radio's (chipset, OS etc.) completely new from zero with only one thing in mind; Make use of the full capabilities embedded in the new ac protocol with one goal only; beat fibre in roll outs.
Their Omni Solution (which are 4 singlepol sectors in fact) is nice designed.
Using Atheros a lot can be done. You see this with epmp and the ubiquities. What I find most promising at the moment is the combination of shielding, antenna design, RF filtering and gps sync you see with the prism station. And a second wireless card just for scanning in background (Great tool). For ptp in 5GHz yo cant beat the Airfibers at the moment and they make an announcement soon ... And guess what is synccompatible ... To bad MT never decided to put more money/people in their wireless hardware department. But this is a business decision I cant appraise.
We stay with MT for routing/MPLS. Wireless is always a mix of vendors as no one is good with every aspect/frequency.
Sent from my ONEPLUS A3003 using Tapatalk
Install a Speedtest.net server at your gateway. At least you avoid (inter-)net and backbone delays. Due the lesser hops ping times are also better.With 12 meg you are fine with MT. Problems start with 20. Customers get this speeds with udp but they complain with tcp. As speedtests are tcp ... I am so bored with this speedtester calls. This problem is for years now and consumed a lot of my time.
This and some towers run out of frequency urged us to move on. We tried epmp but failed due to worse near channel rejection. Had to move on further.
To keep up with bandwidth needs we mix. Short hops we do 60Ghz, mid range is licensed or airfiber. Depends on usage or frequency congestion. Moving backhauls out of 5GHz to make 40Mhz Channels possible. Sync also helps to keep 5GHz usable.
With nv2 and APs with more than 6-10 connected cpes customers *never* see 20MBit/s tcp even when all other cpes are quiet. This is not a simple bandwidth problem it is how nv2 works, Speedtest.net does 4 tcp-streams so you see a bit more than with a single tcp download/ftp. This got some way better with newer versions but still is not good. With all versions newer than 6.35.4 connection algorithm has changed (silently). If there are more than 2 APs with the same SSID the CPE might not select the AP we direct him with access-lists.Install a Speedtest.net server at your gateway. At least you avoid (inter-)net and backbone delays. Due the lesser hops ping times are also better.With 12 meg you are fine with MT. Problems start with 20. Customers get this speeds with udp but they complain with tcp. As speedtests are tcp ... I am so bored with this speedtester calls. This problem is for years now and consumed a lot of my time.
This and some towers run out of frequency urged us to move on. We tried epmp but failed due to worse near channel rejection. Had to move on further.
To keep up with bandwidth needs we mix. Short hops we do 60Ghz, mid range is licensed or airfiber. Depends on usage or frequency congestion. Moving backhauls out of 5GHz to make 40Mhz Channels possible. Sync also helps to keep 5GHz usable.
We've been battered by 'speedtesters' in the past wining about low speeds and high pingtimes which was more due saturated backbone, overused internet or saturated speedtest servers. Now we run our own server (free! You only need a good powerfull server) and our clients now (default-)test to our own speedtest.net server. 90% of the complaints vanished....
Second advantage is that running long(er) bandwidth tests on a P2MP network sees jumps up and down in the speed to the client simply due the fact that others have to be served too. With speedtest its a limited time test which usually give better results.
If we sell 20Mb contracts that doesn't mean the client get 20Mb 24/7. It's maximum and with the speedtest you can show the client he gets what he pays for... almost 20Mb. Off course, when he runs a long running download the average will be lower....
We'd usually try to keep the network not more then 3 or 4 update versions behind so we should be having new developments relatively soon.With nv2 and APs with more than 6-10 connected cpes customers *never* see 20MBit/s tcp even when all other cpes are quiet. This is not a simple bandwidth problem it is how nv2 works, Speedtest.net does 4 tcp-streams so you see a bit more than with a single tcp download/ftp. This got some way better with newer versions but still is not good. With all versions newer than 6.35.4 connection algorithm has changed (silently). If there are more than 2 APs with the same SSID the CPE might not select the AP we direct him with access-lists.
We talked to MT on all of this ...
We design our network so customers see their bandwidth. Of course we oversell but at a rate they usually see their bandwidth.
Is there much involved in setting up a speedtest server?Install a Speedtest.net server at your gateway. At least you avoid (inter-)net and backbone delays. Due the lesser hops ping times are also better.
We've been battered by 'speedtesters' in the past wining about low speeds and high pingtimes which was more due saturated backbone, overused internet or saturated speedtest servers. Now we run our own server (free! You only need a good powerfull server) and our clients now (default-)test to our own speedtest.net server. 90% of the complaints vanished....
Second advantage is that running long(er) bandwidth tests on a P2MP network sees jumps up and down in the speed to the client simply due the fact that others have to be served too. With speedtest its a limited time test which usually give better results.
If we sell 20Mb contracts that doesn't mean the client get 20Mb 24/7. It's maximum and with the speedtest you can show the client he gets what he pays for... almost 20Mb. Off course, when he runs a long running download the average will be lower....
No. You need the hardware (A good server will do.) and a pipe (at least 100Mb simmetric I believe.). All info can be found on their website.Is there much involved in setting up a speedtest server?
Are these actual throughput figures or just from data sheet spec?Mimosa product OS & Hardware like microwave product, all process in the level 1(throughput bypass)
Mikrotik product OS when traffic running processor load too high . If setting fast path when throughput up to 80-90%, processor in 40-50%.
e.g:-
Mimosa 5Ghz-AC-80Mhz & Netmetal 5Ghz-AC-80Mhz
Mimosa throughput can up to 700Mbps half duplex. 350Mbps Full duplex ( TCP ) 950Mbps udp half duplex 550Mbps full duplex. ( distance 20KM )
Netmetal 5 throughput only can up to 400Mbps half duplex, Full duplex up to 200Mbps ( TCP) udp half duplex 500Mbps, full duplex 300Mbps. ( distance 5KM ).
Nv2 normally throughput testing both difference, sometime upload higher sometime download higher. ( can try 820.11 ) this more stable & both way testing is balance.
Mimosa throughput testing both way balance.
ps: price difference. 1 unit mimosa can buy few unit NetMetal.
Ok, let me try to understand what you wrote..Rudy 2x SA distance between 1m protocol 802.11 same channel running better than nv2 sync.
Yep 2 units of SXT-SA5 (SA5-AC units are much stronger don't know, noise floor should be bad) as AP on opposite direction 1 meter apart same channel 802.11 mode work better than in nv2 sync mode.Ok, let me try to understand what you wrote..Rudy 2x SA distance between 1m protocol 802.11 same channel running better than nv2 sync.
2x SA = 2 x SXT-SA ok
So you are saying that instead of 2 SXT-AC's 1 meter apart from eachother work better with 802.11 in the same channel then when they are in NV2 sync mode?
If that would be the case it would mean the NV2 sync mode is a complete failure..... Why even mention it exists if it doesn't work.
Either I don't understand what you tried to say, or the Mikrotik NV2 sync is a complete failure or needs much better explanations.
Simple answer to that. They don't.So I am also wondering on how these Mikrotik guys do it. We tried now 3 different setups with scenarios where sync could save us spectrum but it just won't work.
Each time the two AP's are connected by ethernet to a switch (in one case both on the ports of a PoE Omnitik) for PoE out but everytime in the same /29 or /28 network. So I would presume the sync over the ethernet should have no issue.
Mikrotik? Any suggestions?
I am awaiting Elevate for over a year now.. works fine for some Ubiquity CPE's we'd happen to have but 99% of my network is MT that really needs updated. Cambium is missing a lot of our money now since stil no elevate for MT. My money is now flowing to Mimosa. And it works GREAT!Maybe Elevate will save us the rest of our MT investment, which for myself is a _LOT_ of money, but in the mean time I have stopped deploying MT on ALL new builds and we have replaced a couple hundred RB411 CPEs and a few dozen RB433AH AP's with Mimosa or Cambium and continue to do so.
Zod
Can I ask which is the best Mimosa or Cambium?
.....................
Maybe Elevate will save us the rest of our MT investment, which for myself is a _LOT_ of money, but in the mean time I have stopped deploying MT on ALL new builds and we have replaced a couple hundred RB411 CPEs and a few dozen RB433AH AP's with Mimosa or Cambium and continue to do so.
Zod
Not a lot people will be able to answer that properly. You need to run both systems in a sort of similar setup and environment to make real comparisons.Can I ask which is the best Mimosa or Cambium?
Thanks Rudy, our customers from most high sites are between 5-15+Kms AP to CPE, we use 23dB CPE'e for all customers!Not a lot people will be able to answer that properly. You need to run both systems in a sort of similar setup and environment to make real comparisons.Can I ask which is the best Mimosa or Cambium?
We have 3 AP's now from Mimosa each with 40+ clients all short distance (500Mtr max) in a high used spectrum where we managed to get 80Mhz relatively 'clean' and although we are not in full sync yet the performance already blows Mikrotik miles behind....
We have one ePMP2000 with beamforming and 3 true eCambium antennas and 2 Ubiquitys with the elevate firmware. They all run at at least 1km, most around 2-3km and they do fine too. We give these clients 40 and 50Mbps simmetric and so far haven't had complaints.
We still have to try Mimosa at bigger distance but Mimosa really likes the higher signal levels for good performance. But this performance in top througputts is higher then even the secs of eCambium.
Price wise Mimosa is more interesting since you can 'shop' around for good deals where eCambium has strickt price policy so hardly any discounts to be found.
I made some investment calcs on Mimosa versus eCambium and found a 40 clients network on one AP would make Mimosa the winner since we also need to buy all new CPE's (like on Mimosa too). Maybe when eCambium comes with elavate firmware for Mikrotik that might tip the balance in case of transition of a network.
But a 'real' comparison on performance differences under the same circumstances I cannot give. Both are fine to good to very good and at least much better then MT. (Sorry guys, but that's the way it is... )
One of the problems I was facing was to find reasonable price 23dBi duo pol antennas to use with C5c devices. But I found some (http://wisnt.com/download/Datasheet/ANG5829-data.pdf) reasonably price that we are about to test.Thanks Rudy, our customers from most high sites are between 5-15+Kms AP to CPE, we use 23dB CPE'e for all customers!
Our challenge appears to be somewhat similar to what you have but to add we have also to switch from single polarity which gives us a additional headache;
(1) Do we purchase Mimosa or Cambium CPE's but can they work effectively over 8-18KMs from the AP,
(2) Or simply purchase the higher gain LHG XL HP5 use Mimosa in Interop mode and wait for Mimosa to develop more C range products or for Cambium to finally release "Elevate" so that the LHG XL HP5 could then be used to Cambium AP's.
Nice info. My test are already done that way APs are placed on opposite direction.just adding some info for readers here that might be of interest
Syncing Ap radios is not as simple as "turning it on" , whatever brand you are going to use.
also Mimosa isnt the holy grail if the design of the network isnt build for it.
the have very good documentation on that :
http://ap.help.mimosa.co/gps-sync-spect ... e-gigapops
i presume this wont be very different than Mikrotik radios in NV2 sync mode.
I never played with 80 since I'd presume that's too much. I played with 70, 60 and 50 but really its a bit hard to see any differences. To see the effect of these settings you need to monitor multiple clients with some real time high traffic load to see differences between each of the settings. Otherwise you're doing not so much more then guessing...i noticed in lab test config with NV2 that you have to set the dynamic download to 80 to achieve good download rates to clients, set to 50 was no good.
did you played with that too ? and what where your experiences.
I think a lot of us feel that way.Unfortunately I think they have no intention of developing the wireless leg of their company.
We are quite happy with CCR (in good ambient environment). We like the RB1100AHx4 very much. Fast and very flexible regarding power input. We like Powerbox Pro.Further to stop using CCRs in it's entirety (PSU issues, BGP issues), we too, are seriously considering alternatives (Cambium / Mimosa) to Mikrotik on the wireless side.
Mikrotik's loosing traction fast. Unless v7 is a magic bullet that gets released, very, very soon... I see tough times ahead for MT, even Ubiquity has far better stuff on the market today. We, simply cannot depend on MT any longer, except for the bare basic non-essential stuff.
I agree with CCR and RB1100 being good products!We are quite happy with CCR (in good ambient environment). We like the RB1100AHx4 very much. Fast and very flexible regarding power input. We like Powerbox Pro.
CRS317-1G-16S+RM is a very good product and (while still beta) the new Bridging configuration makes it more usable than the older MT-Switches. And when it does MPLS in Hardware at 10G wirespeed ...
So yes MT wireless falls back. I guess it is used mostly in countries where every $ counts. I guess this is the reason they bring new/old cheap 11n devices. It works reliable but does not scale.
But they still have a lot of products which help WISPs.
So test and select Products for purpose and dont expect one vendor to do it all. Noone does LTE, Licensed wireless, 60GHz, unlicensed wireless, routing, switching, CRM, ...
And dont trust markteting and fanboys. We see Mimosa and Cambium ePMP fall to it's nose with interference where nv2 still work. They dont have a magic bullet. And we soon have 2018 and ePMP is still 11n. We dont buy 11n for a year now ...
I agree, MT is versatile in a lot. We have their routers in our core and they probably will never be replaced. But where we years ago (when tdma became in play) had to make a decision in what hardware line we wher going to follow we'd choose MT in these days since the first UBNT lines where crap to say the least.Noone does LTE, Licensed wireless, 60GHz, unlicensed wireless, routing, switching, CRM, ...
And dont trust markteting and fanboys. We see Mimosa and Cambium ePMP fall to it's nose with interference where nv2 still work. They dont have a magic bullet. And we soon have 2018 and ePMP is still 11n. We dont buy 11n for a year now ...
All true but dont neglect e.g. problems with other gear. We have 3 Metrolinq links running. 60 GHz is good but this equipment has it's own problems. The first batches had to be replaced due to faulty HW. The SW shows some showstoppers now and then and needs to be rebooted. The last update of a V2,5 link makes the remote end reset to factory default. No problem as this is a backup link. But in no way I trust this equipment at current stage. We wait some months before we consider using more of this gear.I agree, MT is versatile in a lot. We have their routers in our core and they probably will never be replaced. But where we years ago (when tdma became in play) had to make a decision in what hardware line we wher going to follow we'd choose MT in these days since the first UBNT lines where crap to say the least.Noone does LTE, Licensed wireless, 60GHz, unlicensed wireless, routing, switching, CRM, ...
And dont trust markteting and fanboys. We see Mimosa and Cambium ePMP fall to it's nose with interference where nv2 still work. They dont have a magic bullet. And we soon have 2018 and ePMP is still 11n. We dont buy 11n for a year now ...
The ROS-versatility-range of products-pricing etc. factors made MT a winner and after the initial issues NV2 started to work fine too where others (mainly Ubiquity in my region) where still having issues.
But then MT's wifi evolution sort of stopped where other moved on and new ones entered the market.
How many years have we seen the 'sync' discussions on this forum. Now it finally sort of cautiously arrived it doesn't seem to work. (We can't get it to work at least....)
The many software updates not always brought 'good', sometimes complete disaster even on 'stable' versions.
Wireless management absolutely is not that ergonomic anymore as in the 'old' days where you could fine tune links without loosing the link after each 'enter' or 'accept' of yet again a small edit.
(I mean; If I just edit the comment field in an AP and hit 'enter' all clients disassociate? That was never like that in the old day. In old version you could even swap the order of the 'connect-to' listing without loosing the link of a CPE. Now you only have to touch anything and the link drops......)
We have one eCambium 2000 AP with beam forming and have 5 high demanding clients on it. 50/50Mbps packages for some of them and it always works and I see it regularly been used.
This in a heavy congested spectrum on a 20Mhz channel width. And the links are stable. Try to do the same with a Netmetal and apart from the link been less stable its almost impossible to surpass the 60-70Mbps aggregated throughput with a handful of clients attached. I have done several tests and best I could ever get in 80Mhz wide channel was 90Mbps (no, not a cable limit).
Some back haul links that really needed to have more then 100Mbps aggregated plus lots of spare have been replaced with Airfibres that run with 300 or 400Mbps aggregated stable connections.
With Netmetals on the same links with the same Jirious antennas, same frequency we tried and tried (nv2, nstream, csma 40Mhz, 80Mhz) but never more then 150-160Mbps aggregated. So we tried the Airfibres and are happy with these. Have 3 links running with them.
Then we replaced two other links for Mimosa B5c's and although not as good as the Airfibre's the still outperform the Netmetal and with the channel spit for ul/dl we'd manage to stay away from most sever interference where with a single Netmetal setup this is just impossible.
Last, but not least. In an heavy used by 5Ghz wifi dense urbanisation we had 6 Netmetals/Omnitiks serving some 150 clients but lots of complaints about poor speeds and broken streams. After a year of struggling we decided to try Mimosa. They came finally (first in Spain!) a year after their first promise and we started to use one. Even in 'interop' (=csma wit RTS/CTS) it worked much better then the previous Mikrotiks. We've had the work previously done by two Omnitiks to server 2 x 25-30 clients (=55 clients) by one A5 that served those 55 plus 10 more (=65 clients!) and speeds were back to were we'd wanted them. Lower latency.
I'v done tests and could easy push the aggregated throughput over the 200Mbps levels. Even when in csma mode with some 'n' SXT's still present in a full working AP network (some 30-40Mb aggregated traffic from clients) I could push my test SXT-ac in 40Mhz channel to 60-70-Mbps. When AO was set to 80Mhz (the legacy units still connected in 40Mhz!) my SXT-ac could go to 180Mbps ! And this where adjacent frequecies are 'loud' and the working channel even 'sees' some remote AP's using the same.
Since some weeks we have now two A5's at only 125 meters separation working in full sync with their tdma mode and 40 on the one (44 is still their limit for tdma) and 6 on the other but they all work flawless. 60-80 and 150Mbps including not a problem for their C5's.
Now the Mimosa system has off course better specs then the Mikrotik equivalents, cost quit a bit more too. But MT just haven't an answer to this. And as an operator I have to keep the fiber boys out of my backyard so had to move on.....
I am also using several 60Ghz links from Metrlinq and although not as good as the specs they do a great job. 500Mbps over 400 meters is not an issue.
I say the Mikrotik 60Ghz platform in Milan a year ago and that is only recently fully available. But up to 100 meters only. Not a lot of use for that. Their bigger units promised by then are since fallen quiet...
And now Metrolinq has an triple (60,5 and 2,4Ghz 12dBi Omni with beamforming in each band that can 'pump 10Gb of aggregated traffic! I need that one to serve my 6 60Ghz links leaving one tower with two 5Ghz AP's and a 2,4Ghz hotspot system.
If this was all Mikrotik it means I can throw away 12 Mikrotik units to be replaced by one Ignitenet unit only!
No, Mikrotik has to come up with beter news or I am afraid in a years time you won't see me a lot on this forum more neither....
This is a MT forum. So I do not want to educate . It is not LTE.What LTU gear ?
OK - I suppose U it's a 8 letter word ending in "i"This is a MT forum. So I do not want to educate . It is not LTE.What LTU gear ?
Well, several other platforms passed the discussions here including their technology and performances....This is a MT forum. So I do not want to educate . It is not LTE.What LTU gear ?
Typical of a sales promo - but its the festive season and "this year" has only 19 days left, so we won't have long to wait for this new technologyhttps://www.fool.com/investing/2017/02/ ... rying.aspx
.."will be rolling out a brand-new technology called LTU this year"..
Purchasable via BetaStore in USA as PTP-Product for 2 months now. Late and beta but no paperware.Typical of a sales promo - but its the festive season and "this year" has only 19 days left, so we won't have long to wait for this new technologyhttps://www.fool.com/investing/2017/02/ ... rying.aspx
.."will be rolling out a brand-new technology called LTU this year"..
1. We tried to do it exactly as prescribed by MT. Tower on a hill. Both sides covered with an sector (and thus in opposite direction). Both sectors are separated by at least 2 meters horizontal and also the steel construction is in between. Ono of the sectors is an SXT-ac with shield. The other a Netmetal on a 'Gold' 60 degrees sector.Ok,
So I have had an actual email discussion with an actual Mikrotik agent. Here is his answer:
Hello,
You can look up in Internet how all the frequency re-use plan works.
The CPE should see only the one of the Synced AP - it can't see the other AP otherwise the connection will not work correctly as there will be collisions.
The best plan to have 4 sectors.
The AP to sync with the other AP that are located back to back on the tower so the CPEs would not see the other AP as it is on the other side of the tower.
Then you can put another AP 90 degrees on different frequency from the other AP and make it back to back to the other AP on the tower.
You can try to repeat that setup putting more such 4 sector combinations on the tower with more frequency spacing so they do not overlap.
So according to this, you can't have sync working on AP's facing the same direction. Which makes the whole idea of sync redundant for me. I mean, in what perfect world are your connections all balanced accross 360 degrees ???!
Without deep knowledge of the chipset features I guess this will never work very good with HW not designed for sync. At least I would expect a hint from MT which exact HW/Chipset they tested/designed this for (I cant believe this works across all Atheros Chips without adaption). Where is the whitepaper describing the test-setup with results? Dont waste your time ...1. We tried to do it exactly as prescribed by MT. Tower on a hill. Both sides covered with an sector (and thus in opposite direction). Both sectors are separated by at least 2 meters horizontal and also the steel construction is in between. Ono of the sectors is an SXT-ac with shield. The other a Netmetal on a 'Gold' 60 degrees sector.Ok,
So I have had an actual email discussion with an actual Mikrotik agent. Here is his answer:
Hello,
You can look up in Internet how all the frequency re-use plan works.
The CPE should see only the one of the Synced AP - it can't see the other AP otherwise the connection will not work correctly as there will be collisions.
The best plan to have 4 sectors.
The AP to sync with the other AP that are located back to back on the tower so the CPEs would not see the other AP as it is on the other side of the tower.
Then you can put another AP 90 degrees on different frequency from the other AP and make it back to back to the other AP on the tower.
You can try to repeat that setup putting more such 4 sector combinations on the tower with more frequency spacing so they do not overlap.
So according to this, you can't have sync working on AP's facing the same direction. Which makes the whole idea of sync redundant for me. I mean, in what perfect world are your connections all balanced accross 360 degrees ???!
Clients on one side of the hill have no way of being able to 'see' clients on the other side. The hill is in-between.
The original setup had separated frequencies on both sectors. We made speed test from both sides and noted the speeds.
After that we picked the best frequency of the two and set both sectors to that one and in full sync according Wiki.
Again we took some clients and did speedtests and noted these.
We already saw the CCQ's dropped on most antenas but worse, the speed to each clients for the test was no more then half. Sometimes even less.
Both sectors both have their own router/radio but in transparant bridge mode. Both are part of the same /28 network and both connect to the same Netonix switch that trunks all AP ports towards one port that has a tower router connected for OSPF, routing etc. Si in Neighbour both sectors and even their clients are visible (by mac only. Clients all have PPP tunnels for authentication towards remote PPPoE server.)
In my opinion this would be the best scenario situation for having two back to back connected sectors working in Mikrotik sync. But it doesn't.....
2. We have another vendor system where we have two Omni directional AP's at only 120 meters apart from eachother and working in full synch (GPS)
We only have to make sure CPE's don't see both in their same working beam. It works fine! Very high MCS rates with high PHY rates.
These have other technology but the sync works....
I think MT has to work harder to get it working. At present it doesn't.
Only by this last line again it seems Mikrotik takes it for granted we users do their tryouts and testing. Old members on this forum that are regularly posting comments and or failures already knew!Ok,
So I have had an actual email discussion with an actual Mikrotik agent. Here is his answer:
Hello,
You can look up in Internet how all the frequency re-use plan works.
The CPE should see only the one of the Synced AP - it can't see the other AP otherwise the connection will not work correctly as there will be collisions.
The best plan to have 4 sectors.
The AP to sync with the other AP that are located back to back on the tower so the CPEs would not see the other AP as it is on the other side of the tower.
Then you can put another AP 90 degrees on different frequency from the other AP and make it back to back to the other AP on the tower.
You can try to repeat that setup putting more such 4 sector combinations on the tower with more frequency spacing so they do not overlap.
+1I guess us moving into Mimosa was the right decision in the end.
Man UBNT / Mimosa / Cambium same as mikrotik u cant use sync this way use brain or read UBNT manual. I have meny AirFiber 5x p2p links rly hard use sync sometimes just cant.Update from Mikrotik:
Hello,
This is how the wireless sync features are working and there is nothing which could be done to improve the situation.
If you have lot of clients pointing in one direction then the only solution is to use different wireless frequency.
I guess us moving into Mimosa was the right decision in the end.
Only part of the truth. Even in the same direction sync helps as it does reduce near channel interference. Of course you cant use the same channel but as wifi-radios are bleeding quite much into the neighboring channels (some vendors take measures to reduce this) you have to separate frequencies quite far depending on TX-Power, channelwidth and antennapattern.Man UBNT / Mimosa / Cambium same as mikrotik u cant use sync this way use brain or read UBNT manual. I have meny AirFiber 5x p2p links rly hard use sync sometimes just cant.Update from Mikrotik:
Hello,
This is how the wireless sync features are working and there is nothing which could be done to improve the situation.
If you have lot of clients pointing in one direction then the only solution is to use different wireless frequency.
I guess us moving into Mimosa was the right decision in the end.
I am running 3 or 4 OmniTik 5 ac now, all on 6.40.5 and fw 3.41 and have no issues...So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
This one had no issues, too. Just enabled access-list and boooom. Looks like the radar event happens as soon as the first cpe connects. Spectrum is quite free there.I am running 3 or 4 OmniTik 5 ac now, all on 6.40.5 and fw 3.41 and have no issues...So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
Is that occurring with wireless set to frequency-mode=superchannel ??So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
Did not test this. Tried to get people online asap.Is that occurring with wireless set to frequency-mode=superchannel ??So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
That setting is the quickest way to get clients registered??Did not test this. Tried to get people online asap.Is that occurring with wireless set to frequency-mode=superchannel ??So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
Ah. Ok. Tested. I thought this needs a licence I have to obtain.That setting is the quickest way to get clients registered??Did not test this. Tried to get people online asap.Is that occurring with wireless set to frequency-mode=superchannel ??So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
Wel, let's put it this way; If everybody in the world is going to work according regulatory domain 802.11 based WISP industry would be marginal....Ah. Ok. Tested. I thought this needs a licence I have to obtain.That setting is the quickest way to get clients registered??Did not test this. Tried to get people online asap.Is that occurring with wireless set to frequency-mode=superchannel ??So another one:
Had to change a wireless setting on an OmnitikAC running 6.40.4 with nv2.
Interface goes into radar detect and finds radar on every single frequency in 5GHz. (Guess what. There is no radar).
Upgrade to 6.40.6 does not help. Downgrade to the latest SW without radar detect = not allowed by ROS.
Disable nv2 for plain 802.11 -> no radar detect
So all connected again. But one station disconnects with extensive data loss
Well, I am not really enthusiastic in putting a rc version on a production network..... The more since I see that NV2 in PtP setup is having serious problems lately with arm devices.@WirelessRudy
Please check mikrotik Sync again with NV2 improvements version 6.42.rc49!!!
I'm waiting for your report
Thanks!
I made several posts in several wireless related treads this week. Search for them and you'll find a lot of info.@WirelessRudy
You have news?