Suburban Internet - Is It Really Feasible


#1

I’m not sure where to put this, but from the descriptions, this sounds like the place. I would just like to start a discussion about how feasible a suburban MicroPOP business truly is, given availability of high-speed cable. I know Rory Conaway does this - his Tales from the Towers series is great - and Sail Internet is another. It does seem like a rather daunting endeavor. Using residential housing for the POP locations - how are the experiences with that? Needing access when there’s an outage for example.

I know the pitch - unbundle, cable keeps going up, nobody really needs a gigabit at their home. I’m just wondering how the logistics of it is working out.

Thanks,

Kevin


#2

It works. Were doing the MicroPOPs in the small town of Ravenna, NE. I have over 300 C5s connected to A5s and scattered across town. I opted to go microPOP with the design in order to provide clients with fiber fast internet for a low price.


#3

I am also interested in this. As Kevin says, the business case is understood and Mimosa have done some nice videos around that, plus the Triad/Sail/XL/etc stories. So a +1 from me for any real-life experiences around logistics - pop’s on residential properties, performance/reliability/stability at 80MHz in the (sub)urbs, etc.

Rich


#4

Thank you Kent. Would you be willing to share some more details? Overlap with Richard’s post.

  1. Installing POPs on homes. You need the AP and a couple of backhauls, so do you have any install pictures?
  2. How about service/maintenance issues at the homes?
  3. 80Mhz channels that are required for high speeds. How successful have you been in provisioning

My area is much more densely populated than Ravenna, so I’m worried about 3. I’m going to do some site surveys this weekend. Also, since I’ve been out of the WISP business for some time, IPv4 is going to be an issue. I’d like to provide a G2 for the install, but from what I can tell, they don’t offer IPv6 support, and I would like to start dual stack. I can get a 1Gb connection to the initial service area for about 1K/month, including roof rent. Economically, this is doable, but the devil is in the details, as always!

Thanks,

Kevin


#5

Hi Kevin,

I have 14 A5s scattered across town. 9 A5s on the West side and 5 on the East. Im using a variety of different brands/models for backhauls. The end goal is to have all the backhauls on either 23,24,60, or 70/80GHz. We currently have 9 60GHz links and the rest are fed with B5-Lites. I plan to replace the remaining B5 Lites with higher freq backhauls spring-summer 2017 to further clean up 5GHz spectrum.

All of the sites have a case mounted outside with battery backups so its accessible for us to service 24/7. The electrical is hardwired into the building or homes electrical so we don’t have to worry about someone unplugging our equipment.

I had to deploy so many A5s due to all of trees in town. All of the backhauls are shooting under the trees. I wanted to get the highest PHY possible on every radio so I opted to deploy a lot of A5s. Six of the A5’s are mounted on 50 ft or less Rohn 25 towers the remaining A5s are on home rooftops. All of the A5s are running in 40 or 80MHz channels. I was running 80MHz channels on all of the A5s until the trees lost their leaves. Once GPS sync is available I will be able to run 80s again. We haven’t had any service/maintenance issues since these were deployed. We did a regular five year lease with every person that has one of these micro pops.

We have had 100% success with the A5s. Every customer we install can pull 200-400Mbps download.
A lot of work, yes… but It was far cheaper than deploying LTE and much, much faster.

Below are pics of a few of the sites…


#6

Lol I love the tree install.


#7

Kent, you mentioned that you had to use a lot of A5s due to the town having many trees.

What kind of A5-14 and A5-18 coverage are you seeing in areas with high tree density?

Are you using C5s as CPE? Or are your customers pulling 200-400Mbps without it?

For your roof-mounted A5s, are you seeing significant signal loss from roof & wall structures?


#8

Thanks Kent! I see some MetroLinqs in those pictures. I’ve heard reports of lockups. I contacted IgniteNet, and they said they were expanding development to put more resources on that issue. Wondering if you’ve seen that.

I’ve been working on a kit for these POPs. Current list is something like this. Lots of Tycon, but that’s what I know at the moment.

  1. Enclosure - Tycon ENC-ST with power vent. I’ve used smaller Hammonds in the past. Need space for battery this time
  2. Smart battery charger - Tycon BC24
  3. Tycon Remote Monitor and Control
  4. Edgerouter X. I want routed POPs, and this one is POE powered. Maybe MT, but I’m familiar with Vyatta, and haven’t used MT in years
  5. Netonix WS-8-150-DC switch for POE.
  6. (2) IgniteNet 60 Gig backhauls
  7. A5-14 AP
  8. Battery. Not sure of capacity required yet.

I suppose I could just run a UPS and power everything off AC - eliminating the charge controller and Netonix - but I like the idea of all DC powered gear.

Now, the question of Wireless Routers. As I said earlier, since I’m getting back into the WISP game, I don’t have ipv4 addresses, and I’m sure I won’t get many from my provider. So I need to start dual-stacked, I think. That might imply providing customer routers, as I would have to ensure what they have at their home is v6 capable. Still chewing on that one.

Nice conversation we have going here - thanks everyone for the input!

Kevin


#9

@riesling Im able to still achieve good PHY going thru a light amount of foliage. The trees are very thick and are all mostly deciduous. We were using UBNT radios at first but have upgraded all the clients to C5s.

No signal loss from roof & wall structures.


#10

@Kevin3 Oh yes, the IgniteNets… I have seen that issue a lot unfortunately. I didn’t have any issues with them until cold weather hit. Then the lockups became very frequent. They sent someone to Ravenna and they replaced the bad wireless cards that were causing the lockups. They also gave me some beta firmware that has fixed my issues so far.

Your list looks good… Im using a much cheaper enclosure. Not as big but far less expensive. Im using these: https://www.polycase.com/yh-161407
They don’t have any ventilation as is but the site does have vents you can buy. Im using Cisco switches. Have not had any issues with the switches being in those cases and getting to hot.

I would check with your provider. If its one of the larger carriers they generally have IPv4 addresses you can use for a price.


#11

@Kent Thanks! What I’m trying to understand is the effective radius of coverage for the A5-14s and A5-18s under different foliage scenarios. Something like this (all numbers = made-up examples):

                      A5-14 to C5 PHY     A5-18 to C5 PHY

Light foliage
@ 600 meters 370 Mbps 430 Mbps

Light foliage
@ 1000 meters 290 Mbps 350 Mbps

Medium foliage
@600 meters 310 Mbps 380 Mbps

etc… (these can be either estimated rates or precisely measured rates)


#12

Thanks Kent. Really nice use-case write-up. This is new ground for a lot of us and it’s really good to get some real-life feedback that’s not part of Mimosa’s marketing collateral.

I’m curious as to what the uptake is like and whether you are running against stiff competition (cable/fibre/dsl @ what speed)? I can’t quite tell, but by the looks of things in the pictures you have an A5 for quite small clusters of homes (20-30?) and those installs (A5+MetroLinq+Cisco) can’t be cheap.

I think there is some question and challenges around AP hosting on customer premises that no-one yet knows the answers to, simply because this is a new model and it hasn’t been encountered. Like if the customer moves house, or they no-longer want to host the antenna, or they default on mortgage, or die, etc…

I’m personally looking at this for a more dense sub-urban environment - ~1,000 dwellings per km2 where I am facing fibre competition, albeit expensive in comparison. My plan is to have 2 or 3 close neighbours all hosting AP’s so I can load balance, and have customers fail-over to another adjacent AP should one go down, or the host wants it removed. But can’t even think of that until GPS comes along.

@riesling - have you seen the planner on cloud.mimosa.co? It’s no substitute for real experience, but it has a PtMP planning mode where it does give some idea of the A5’s range given different attenuation scenarios.

Rich


#13

I would get a /26, but beyond that I’d have to purchase and they keep going up. I’d like to be ipv6 ready. What are you using for Wireless routers for your customers? Do you provide or let them purchase whatever they want. That’s one issue with going ipv6 - you have to make sure the router supports it.

I’ll check Polycase. I’m going to need forced ventilation out here.

Thanks,

Kevin


#14

Do you have to give every user a public IP? We have a /27 but run a private network for all subscribers. We 1:1 NAT a static public for the few that need one. For those that do not, we assign a different public to each VLAN and each AP gets its own VLAN for customer access. As a result, we have ~40 users each sharing a public. Works very well.


#15

@Chadwick,

I don’t. Just wanted to avoid double NAT if possible, but that sounds like a good alternative.

Thanks!


#16

Kent, this was a very helpful post! I’d love to get your feedback on a project which we are working on. If you’re up for it, my email is mpool@myontv.com. Thanks again for this post.


#17

Thanks Richard. The cost per MicroPOP for me is $2200. Ive been buying used refurb Cisco switches to help keep costs low. The switch runs me $100 for a 2960G Compact switch.

Im doing about 20-30 homes per A5. There are a couple A5s with 40-50.

I have not started to advertise the new speed plans. Were competing against Charter which is 60/5 for $34.95. I have not decided what our prices will be but I plan to offer up to 150Mbps download speeds. There is a fiber to the home ISP that received millions of gov funding coming to town possibly next year so I want to be prepared for that. They are selling 100Mbps/100Mbps for $65.


#18

Nice. Yeah, the Cisco gear is pretty rock solid so it is a good choice if you can get it for a decent price. I just had to upgrade a bunch of 2950’s, some that had an uptime of 8+ years in the show ver!

I’ll share my experiences vs fibre once I get my first MicroPOP installed, which I hope to get over the holidays or early new year. Our incumbent does 100/20 for around $65 also. The flip side is that it makes backhaul easier.

Rich


#19

I don’t! Poor thing! haha. But what about wind? And growth? We uses a tree years ago but every season and after each storm we needed to re-align the backhauls…

Some of the other pics are also showing some wind sensitivity setups. I’ll guess not a lot of gusty weahter over there?


#20

Ok, we are in the proces of migrating a Mikrotik based network to an A5-14.
Currently we have 36 SXT’s connected (802.11n) range 40-500 meters.

On a single SXT-ac or one of our C5’s (for testing) we manage to get just over 100Mbps download but this is inmediately effecting running downloads of the other SXT’s.

When we only have one C5 doing the download test we can get it to download 120Mb in a 40Mhz channel but as soon as any other device starts to download the overall througput on the AP falls back to some 80-90Mb which off course can be seen in each of the test devices. The C5 falls back to some 40-50Mb.

I am wondering since we are in interop mode the rts cts protocol prevents the network to transport more due collisions? Any suggestions.
@Kent wrote he started with ubnt but dropped that soon. Reason?

Kent also wrote about a whole bunch of A5’s in relative close ranges (urban). But since GPS sync is not working yet how are you deviding the bands? Especial when using 80Mhz you soon run into issues…
And probably you have no other WISP in the 5Ghz around? We are here in Spain in a spectrum full of users which is probably an issue.

You also wrote about some client could drag 150Mbps in… but with 2 or 3 clients silmultaniously? How is that going?

We are dying to get this working to beat the crap out everybody else but so far the A5 is doing better then a Mikrotik Netmetal on a dual pol. antenna but its not half as good as the brochure and you guys are trying to make us believe what is possible…

Can you enlight us with some fine tuning tips? What are typical signal levels on both ends?