Monthly Archives: February 2016

Equipment Box

One of the tricky parts of creating the dbiua network was getting the relay points setup, with a small POE switch, and battery backup.

We initially used Tycon Power Systems UPS-DC1224-9 system.  This was a great little box that had a 9AH battery and a charge controller, and just enough room to get a Ubiquity Toughswitch5 in there.

15186747091_c971612d51_z.jpg

But the 4 allenhead screws were not the best and after being out in the weather for a while, they would break.  So, we switched things out and used all the same parts (9AH battery, TP-SCPOE-1214 charge controller, and toughswitch), but in a Hana Wireless NEMA 12x10x5 fiberglass box.  But, the biggest problem was mounting all the stuff in the Hana box efficiently.

What I ended up doing was designing a bracket to hold the battery up at the top, and then 2 brackets to hold the toughswitch above the charge controller.  I 3D printed both of these things (the blue things in the picture below).

IMG_20150906_165044.jpg

I have uploaded the design of both of these brackets to Shapeways, so if you want them for your own project and do not have a 3d printer, you can just order them from Shapeways.

https://www.shapeways.com/shops/dbiua

You will need 2 toughswitch9 and 1 batterybox12.  The prices for both of these are not marked up at all.

One last part that you need is a POE splitter (Tycon Power Systems POE-SPLT-S
Passive Splitter 5.5/2.1mm DC Shielded): https://www.streakwave.com/itemdesc.asp?ic=POE-SPLT-S&eq=&Tp=&o1=0

This allows you to pull the power off the charge controller POE out port, and power the toughswitch.  In the picture above you can see one end of this splitter plugged into the first port of the switch.  The rest of the splitter is under the switch.

Our latest version of this equipment box includes a RaspberryPI (designed by Brett Marl), that monitors the power of the battery, as well as doing internal speed tests on our network.

IMG_20151128_151042 (1).jpg

 

Radios we use

In a previous post I mentioned the wifi routers we install in members homes.  These are the last part of the network.  So what about everything else?

14743753480_b0268c45d5_k.jpg

We use a variety of radios depending on the situation.  For PTP links we generally use PowerbridgeM5 on both ends, though sometimes we have a PowerbridgeM5 on the upstream side, and a NanoStationM5 on the downstream side.

For one PTP link from the water tank, we have a RocketM5 with 30dBi 2′ dish, and a NanoStationM5 on the other end.  The NanoStationM5 is up at the top of a tree, and sways around a bit.

For our PTMP (Point to Multipoint) links, we always have a RocketM5 with either a 16 or 19 dBi sector, or sometimes a 13dBi omni antenna.  On the downstream side we use either NanoStationM5, or NanoLocoM5, or NanoBeamM5-16, or sometimes for a long link, a NanoBridgeM5-22.  The NanoBeamM5-16 are really great little radios as they have a nice integrated mounting bracket.

15116744199_4e995f5ab8_z.jpg

Some of our links use 900mhz, and for these we will use a RocketM900, with a 13dBi sector (what I call the water heater, because the thing is huge).  Then on the downstream side we have NanoLocoM900.  Sometimes we will use a NanoLocoM900 on both sides of the link if the link is not going very far, or there is just one client radio downstream.

13366906494_f9493b4690_k.jpg

In one location we are using 3.65 Ghz (which requires an FCC license).  Here we have a RocketM365 with a 12dBi Omni, and then NanoStationM365 for downstream radios.

And in another location we are using 2.4Ghz, and a RocketM2 with a 16dBi sector and NanoStationM2 downstream.

These are all part of Ubiquity’s AirMax line.  We are not using an AC radios yet, but we might switch to some of these on our backhaul links in the future.

My Favorite Ubiquity Wifi Router

Our standard member install usually (always), has one radio outside the house.  This is some form of their airmax line.  We sometimes also provide a 2.4ghz wifi router for the member inside the house.  This is not always required if they already have a wifi router.

In order to make things easy, we decided to standardize on a Ubiquity product here as well.

In the very beginning, we used their basic AirRouter.

AirRouter.jpg

We then started using the AirRouter-HP, which has a better range, and is powered by POE.  it is a little more expensive.

AirRouter-HP .jpg

Then I ran across their AirGateway, which is this tiny little thing that plugs directly into the POE brick.  They also had an AirGateway-LR, which was similar to the AirRouter-HP.

airGateway.JPGAirGateway-LRA.jpg

After doing some installs with these, I decided I liked the AirRouters better, specifically the AirRouter-HP.  Why?  Not totally sure, but they seem more tried and true, and you have the ability to plug more than one thing into them if needed.

An Un throttled experiment

Probably all Internet services give people a choice of either speed or data caps.

The DBIUA decided to see what would happen without imposing either of these on its members.

We got the fastest upstream connection that we could, and then we built our wireless network out and tried to provide the fastest speeds possible in an affordable manner.

In order to test the speeds on our internal network, we installed a speed test mini webpage on a server at the water tank.  This allows people to test speeds to the tank.

If you are one or two hops from the tank, you can probably get upload and download speeds in the 40+mpbs range.  If you are several hops away, then that tends to drop to the 20’s.  And some places it’s around 10.  But, that’s a far cry from the 1.5mbps that you sometimes got with DSL.

So, what does our overall bandwidth usage look like going out of the tank to the internet?

ciscotraffic

This is a snapshot of 2 days.  Notice the spike in the evenings.  So even though people may be able to download faster, the reality is they don’t consume that much, and it’s only generally in the evening.  Here is another graph that shows this same traffic over a weeks time.

ciscoweek

So, what about individual usage?  Not all ISP’s graph this data, but we decided to do this so we could manage the network and identify any issues.  I think those ISP’s that might do this, would probably not share this information, because it shows we actually use WAY less than we think we do.

Here is the usage graph for the connection from my house for that same 2 day period:

sutton

Notice the scale on the left.  3.0.  Not 30.0 like the above graph.  And the usage is way less.  The larger blob at the far right is watching some video.  The little spikes are various downloads.  General web browsing, or youtube at lower rez is the other green blips.

Here is another person, with a different usage pattern:

movieusage

Notice the scale change again.  This is someone streaming something high def (the large green blobs).  But, there are still large swaths of time when nothing is happening.

Here are a few more

webusage

moremovie

All of these individual usage patterns flow together to create the one at the top.  We have not had to throttle anyone, or impose data caps.  We allow everyone to use what is available at any given time on the network.

There have been times when a lot of people were streaming something around the same time, and guess what happened?  Things slowed down a little bit for everyone.  Sometimes there may have been a little buffering, but in general it has not been a problem.

The million dollar question is how much speed do you really need?

In my opinion, if you have a reliable 2-3mbps available to you, that is plenty.  If you can burst to faster speeds as needed, that is an extra bonus.  6-8mbps means you can stream very high def video.  But using 50 or 100mbps for long periods of time is actually not very common.

And, personally I have had times when slow speeds are not on our end, but instead on the other end of the connection, at the data center side, where a webserver might be throttled, or on a slow connection.

So, hopefully this gives you all some good real world information about our little socialist network experiment 🙂

Growing and costs going down!

When we launched the DBIUA, in order to cover our upstream bandwidth costs, and to pay down our 3 year loan for our capital equipment, we needed 23 people to commit to paying $75/month.  We got those “early adopters” as Chris Brems likes to say, and then over the past year+ we have slowly added more and more members.

Today, we have 61 connections, with 10 on the waiting list to be installed.

We have almost 3 times the number of members needed to pay for running the system.  What are we doing with all that money?

We had to spend some money on backbone infrastructure to service more people.  And we have installed a backup link at the water tank, so our upstream bandwidth costs have increased a bit.

We purchased backup equipment to have on the shelf in case something bad happens (like a tree falls down).

We are paying off our 3 year loan a year early.

And….we are lowering our monthly rates.  Just like we said we would.

After talking through some ideas, we (the board) came up with the following, which we thought was fair to those who have been with us from the start.

Your first 18 months on the system costs $75/month.  Your next 6 months are $65/month, then 6 months at $55, then after that $45/month.

Our first round of members will start seeing their bills go down in the next month or so.

Even with this tiered pricing model, our projections still say we will end up with a lot of cash in the bank, so we are looking at other capital credit payments to members, along with continued investment in our network.