Satellite Internet Providers 336
pitchblende writes "Our company works in remote locations in Northern Canada. We have been experiencing major communications problems with our current satellite service. We use satellite systems that go for about $1000 apiece, with $100/month in fees. The service is 'shared' rather than dedicated, and our VOIP, etc, has been getting worse by the day lately. From what I can tell, dedicated systems go for $30k and up.
I hope someone(s) out there has some suggestions, recommendations?"
All Satellite Internet Providers are Shared (Score:1, Informative)
That is the dirty little secret of the industry.
And if some punk on your satellite is doing p2p transfers all day, there is nothing you can do to stop it.
If you want a real dedicated connection, you will have to launch your own.
Skycasters has speeds in which we commit data can (Score:4, Informative)
Skycasters has speeds in which commit data can be transferred and they have Platinum Service Plan will be optimized for VoIP
All plans include 1 publicly routable static IP address.
http://www.skycasters.com/broadband-satellite-compare/compare.html [skycasters.com]
Re:Get some of those BUDs in that other thread (Score:4, Informative)
...and the electricity would come from?
Go look at a map of northern Canada.
Terrestrial Wireless (Score:5, Informative)
In order to avoid satellite providers altogether, a number of areas in southern Alberta have made the switch to terrestrial wireless systems. These systems typically operate in the 900 MHz or 2.4 GHz band, and provide each client with a highly directional radio frequency line of sight (it works through trees and bush) to the provider tower, which can be several kilometers away. These systems are very reliable, and boast latency and bandwidth similar to modern cable networks. Most providers do have a bandwidth cap in place, but they are not nearly as absurd as satellite provider caps. Best of all, they cost a fraction of a satellite connection, and the equipment itself costs less than $100 at the client site.
With regards to specific technologies, check out the offerings of Motorola in their Canopy line of products. I'm sure there are many others, but I have experience with this one =)
Re:All Satellite Internet Providers are Shared (Score:2, Informative)
Actually that's not true. You can pay for a guaranteed chunk of bandwidth - your own dedicated PID. And yes you will pay a sizeable amount (5k+ per month for 1-2Mbits).
Re:Get some of those BUDs in that other thread (Score:4, Informative)
solar panel, battery and a small wind genny should do just fine. That's how most weather stations 'up north' are being powered and it would work quite well for a small low power router. There is a Canadian company in the rockies that makes really nice hardware for just that purpose, check out valemount networks http://www.staros.com/ [staros.com]
here is another example.
http://www.ecofriend.org/entry/solar-powered-wireless-router-to-bring-internet-access-to-remote-areas/ [ecofriend.org]
Re:I'm no expert (Score:5, Informative)
I work for a company that has 10+ satelite links and some are better than others. AFIK all satelite operators in Canada use Telesat's satelites, so it doesn't really matter if you switch providers, as you will still be talking to the same bird in the sky.
We use Infosat Communications for our satelite sites, and lately they have been having issues. Their uplink facility is in downtown Calgary and when a storm rolls through (which they have daily now for several weeks) there is a good chance that the uplink facility will lose connection to the bird, and ALL sites will go offline. Outages are usually breif, but a MAJOR pain in my ass.
Once service is restored, likely one of the sites will not come up correctly and I have to call the site and do some rebooting tickery to bring it back online, which SUCKS as most of the people up north can barely tie their shoelaces, let alone work satellite equipment.
We have two different types of satelite dishes. The more reliable of the two (by quite a large margin) is a dish mounted to a 4" pole sunk into concrete. That baby ain't `goin nowhere, and generally works pretty good (but HIGH latency). The other dishes we have are auto aiming, so that, in theory, you can drop the thing anywhere, press some buttons and away it goes. In reality, they can find the satelite in the sky quite well, but if for whatever reason, that connection gets lost, it will not reaquire. Someone has to go out to the site, and play with the equipment. Then when it doesn't come up, we call Infosat, and they get the person on site to play with the equipment, before finally sending a tech.
When one of my auto-aligning dishes goes down, I curse. Usually it takes DAYS to get it back online. I have to get someone on site, then get infosat on the phone...
Anyways, I feel the submitter's pain, as I live with it too. Unfortunately I think you are SOL and will have to live with it, as cellular data can be spotty too (and is unavailable pretty much everywhere north, except northern Alberta. We looked into cellular data and they couldn't/wouldn't give us a SLA so we are still on satelitte.
Re:Amazing (Score:4, Informative)
Residential internet is shared. With satellite, it's the same transmitter for a lot of people. With cable, your neighborhood is on the same cable. With DSL, you may have a dedicated line to the CO, but you're sharing bandwidth at the link the CO has with the rest of the world. Sharing bandwidth is actually a good compromise as it reduces the cost of making sure they have provisions for bandwidth that most people aren't using anyway.
Hosed, Eh? (Score:5, Informative)
We [fmars2007.org] used NetKaster's commercial grade service at 75+degN. The farthest north it has been deployed according to their tech as of last summer. All satellite is shared, but we had good luck with VOIP and even some video conferencing when the weather cooperated. That far north you have to shoot through a lot of atmosphere to hit the bird. I would say if transport size is not an issue, go with two of their 1m dish systems and load balance. That should get you want you want.
Re:Move (Score:2, Informative)
In other words, you're a fucking idiot.
Re:Have you tried a Riverbed device? (Score:5, Informative)
Advice from an "expert". (Score:5, Informative)
I'm currently employed by a US-based VSAT provider and I'm the guy in charge of the IP sections and a good portion of the RF section too. Here's my advice and words of wisdom.
1) You'll get what you pay for. Satellite spectrum *is* expensive, so if you're only paying $100/mo for service, you're being oversubscribed to hell and back.
2) Consumer satellite providers mostly share bandwidth by TDM access. They have a large carrier from their earth station that runs all the time, but your transmitter bursts in a duty cycle set by the system controller at the earth station. Its great for downloads, but it sucks for VOIP.
3) The people who say "VOIP won't work over satellite" are dead wrong. It works just fine. We have many customers in the US and several in Europe that use VOIP just fine. However, they're on "dedicated bandwidth", so there's no TDMing. If they're buying 512kbps of bandwidth, they have 512kbps of bandwidth. But they also pay more for that.
4) I don't know exactly how much data and voice you need, but consider BGAN as a possible solution.
5) And, shameless plug, feel free to contact me and we can see what we can do for you.
Re:VoIP on satellite IP? (Score:4, Informative)
That's some pretty impressive voodoo [google.com].
Re:Move (Score:3, Informative)
I know that some satellite phone systems, namely Iridium and Globalstar use low earth satellites. But I had no idea that geosynchronous satellite phone systems such as Immarsat, ACeS, Thuraya, and MSAT don't actually exist. Thank you for correcting my horrible misconception, Mr. AC!
Re:Get some of those BUDs in that other thread (Score:5, Informative)
Building 10-15 125 mile links with 12' dishes is no trivial (or inexpensive) task when you consider the site acquisition and civil work to pull it off. The operational costs to maintain it in such a harsh environment aren't trivial either. And using 802.11G for this is a joke, and 10-15 125 mile links are going to have an availability that's horribly low, probably in the 70 or 80 percent range. FCC Part 101 (or whatever the equivalent is in Canada) licensed microwave is clearly the way go if they want any reasonable amount of bandwidth and availability, but the cost of this network will dwarf whatever monthly recurring they're paying now.
Pulling off an interesting wireless experiment with hacked and overextended hardware is a lot different than building a production network.
Re:Amazing (Score:5, Informative)
WTF? God only knows how big the Earth is? The radius of the Earth is only ~6400 km. Which is small compared to the 35786 km distance to geostationary. The north pole is the farthest you could be from them on the Earth surface and so it's 36353 from geostationary orbit. So for all your smart assed "God only knows" bullshit, it's 1.5% farther if you aren't on the equator. Moron.
Re:VoIP on satellite IP? (Score:2, Informative)
Actually, that is a long way from the truth. I work with a company heavily involved in VOIP over satellite with customers in Northern Canada. And yes it works fine, but not at $100 a month. Here is the kicker. The Service he is talking about is a home grade consumer product. It was built for Momma and Poppa out on the farm to send the odd email back to the kids in the shiny big city. Sadly every Dick and Jane exeuctive out there decided one of these would be great for the holiday cottage, hence the whole system is oversubscribed...
So a system that is not designed to work with VOIP and over subscribtion.. not so good.
Now, if you are willing to lay down a little bit more moola things change quickly. At under $1000 a month you can get QoS on VOIP as well as dedicated data rates on a shared network. The voice works and works well. We even run it through a variety of call quality monitoring tools and the end result it the voice we hand back to the PSTN is generally better than we get from them!. Yes, there is lag, but if you compare the lag to the average Cell phone call you wouldn't notice the difference.
Latency really is not the issue here, consistency of latency is. If you get a steady 560ms ping, it will work great.
Just don't try gaming on it.
Good, Cheap, Fast. (Score:4, Informative)
Pick 2.
We operate over 50 sites North of 60, with our own uplink facility in Southern Canada. I can tell you satellite bandwidth is EXPENSIVE! Face it, it costs a lot of money to get those things floating around the planet and keep them up there.
There are a few options available to lower the cost, which tend to lead to the 3 options.
The one thing that Infosat does is use KU band. This is cheaper because is has a much greater suck factor. The main problem is the impact of rain-fade. This becomes especially significant in the high north becuase, being on the edge of the footprint (lower gain) and having much more atmosphere to pass through, plus lower elevation angles on the antenna lead to higher noise from terrestrial radiation. (we used infosat links for a number of years, and had the same problems you talk about) :)
The other option is to use C-Band. It has a better footprint in the north and rain fade is a fairly negligible factor. It is also significantly more expensive.
Telesat now has KA band as well, but I'm not that familiar with how it performs, or is priced. There are some issues with KA band, as well because it uses spot beams, so you cannot have a direct link between the East and West without a downlink in the middle somewhere.
Most providers that are offering a "cheap" solution will also provide shared bandwidth solutions. This works well up to a point, as it allows you to make use of extra space when other users aren't using it (commonly called "burst" speeds), but the main problems with this are that everyone tends to want to use it at the same time (more or less) and it is easy for the provider to oversubscribe the link. You may be able to talk to your provider about traffic shaping options to see if they can prioritize voip, although I suppose that is not a very PC remark these days
If you use the service mostly within the communities, there is a last mile broadband solution available for most places in the Northwest Territories and Nunavut (www.qiniq.com / www.airware.ca) using MCS (Clearwire).
Re:Get some of those BUDs in that other thread (Score:2, Informative)
Solar panels? Someone has never worked the high arctic. There are 2 seasons, day and night. In July at the top of Baffin Island there is 24 hour sunshine and in November you have 24 hour nights.
Then depending on location there is the task of anchoring anything you set up. In Pangirtung on Baffin, the airport is secured with steel cables anchored to the baserock. In the 70's watched a DC3 while taxing get airborne as the wind caught it. The wind across the Davis Straits can put a basic hurricane to shame.
VSAT (Score:5, Informative)
I am currently deployed to Iraq. Dues to the specialty of my job, my 3 man team rates our own VSAT uplink. The system we use is made by GCS and is their Cheetah model. Not 100% sure whose birds we use, but I believe Intelsat. This system in general works pretty good, auto acquire dish, integrated router, VOIP, etc. Since it is military, it also provides an uplink to SIPR, also with VOIP capability. The system works alright, but it has been known to be quite finicky, particularly with power sources. While the system is allegedly rated for a wide range, both AC and DC sources, in reality, it sometimes has a problem with generator AC power. Because of my remote location, generators are all there is for power, and anyone who has lived off of generators for an extended period of time can tell you that the power isn't always steady. In the past, power outages due to generator outages have killed the system, requiring one or more components to be replaced. Bandwidth is so so but it does do VOIP fine. Only other gripe I have with it is the management of it. Have to jump through a few hoops to connect to slashdot (something about non-work related . . .).
Re:I'm no expert (Score:2, Informative)
Re:Get some of those BUDs in that other thread (Score:4, Informative)
Indeed, I haven't lived/worked in the high arctic, but the original poster mentioned 'remote locations in Northern Canada', which is not the same as the 'high arctic', and since I've actually lived in such a location for three years in a solar and wind powered home I think I have some relevant knowledge. In fact, in the Canadian winter when it's clear solar panels will have amazing output because they are kept nicely cool by the surrounding air :)
The simple fact that they use satellite there right now according to the OP, indicates they are not in the high arctic or even near it, because there are no satellites in line-of-sight doing telco in the high arctic as far as I know, *maybe* inmarsat has some coverage there (it will cost you dearly though, but if you need them they are almost always the only player), but certainly not at the rates the OP quoted.
Re:Amazing (Score:2, Informative)
VoIP over sattelite? 2-3 second ping round trips? Real time anything over a sattelite is simply not a good idea... The travel distance from ground base to ground base is simply too far to use with direct signals, and we don't yet have the technology to use laser light for long distance communication.
Get some fiber run out to your location. Likely, it will be less than 30K, especially if some others nearby can chip in on a shared GBps link. (I'm assuming you're not more than 30KM from somewhere that DOES have a hard line to the backbone. If not, MOVE.
Re:Amazing (Score:3, Informative)
Not significantly more distance overall. But the satellites lie lower over the horizon, making the path length through the atmosphere (and signal attenuation) greater.
Re:Amazing (Score:4, Informative)
You sir, are the moron.
Pythagoras' theorem is used as follows:
c = sqrt(a^2 + b^2) (a=distance from centre of earth to north pole, b = distance from centre of earth to geo stationary sat.)
c = sqrt(6400^2 + (6400+35786)^2 )
c = 42668.707
That equates to almost 20% more at the north pole, not 1.5%.
Re:Amazing (Score:5, Informative)
You've never looked at a map of Northern Canada, have you?
People who are doing work there are usually there for a good (at least to them) reason, so saying someone should move if their location is more than 30km from a backbone connection is asinine.
Great if you can work in the limitations. (Score:3, Informative)
Latency is usually considered the biggest issue with IP over satellite. The best latency you can possibly get is 550 msec round-trip. If you are working with 2 satellite-enabled sites, your best will be 1100 msec, as there isn't any method of routing packets on the satellite. Packets will have to be sent to the groundstation and be rerouted, re-encrypted and repackaged for transmission to the other endpoint.
The other big killer for satellite IP is the issue of jitter. If you are close to noise floor for receiving or transmitting, you will get a *lot* of jitter as you miss your timing slots or the SIT requests retransmission of the packets. You will also get jitter if you are close to the throttling limits that the provider has enforced in the background that will delay the transmission of frames as a crude QOS system.
Latency kills applications that use lots of small packets for data transmission, e.g. RPC, older implementations of remote desktop, certain VPN solutions. Jitter on the other hand kills things like VOIP that function best with an expected and consistent timing of packet arrival.
The usual method of IP over satellite that I saw in practice was the DVB-RCS protocol http://en.wikipedia.org/wiki/DVB-RCS [wikipedia.org] which in essence packages an encrypted datastream in the mjpeg frames that would be handled as part of a television feed. Knowing the limitations of exactly how the data is transported can go a long way towards explaining the reasons why some apps work great for some people and other apps plainly suck.
There is not a huge amount of bandwidth available on the transponders, and the cost of the use of a transponder and the associated equipment at a groundstation can be frightening.
The issue of pointing accuracy and available power is also critical with satellite IP. The receive strength is important, but not as critical as the pointing required for the transmission side of things. The usual method that we had for pointing was to contact teh upstream provider that had the oscilloscopes on the feed, setting a carrier wave on the satmodem, and changing the point until there was a power peak. Then the antenna was tightened up and comissioning was completed once the routing was set up.
Satellite is good if you work within its limitations. It'll give you good service if your equipment is correctly specified and performs to its spec. It's unfortunate that the cost is so high, but that's the cost of using a transponder on a commercial satellite.
Re:Advice from an "expert". (Score:4, Informative)
Let's put some numbers on it: bulk satellite transponder time is going to cost you $100 to $200 per hour depending on band and desirability of satellite.
If you go with DVB-S2 on 36 MHz transponder Ku satellite, I'm assuming you can generally get by with 8PSK 2/3 modulation (except when it rains very hard), about 7.1 dB Es/N0, for a data rate of ~60 Mbps.
To just break even on $100/month service, you need to revenue of around $100k a month (add in uplink equipment depreciation, internet access costs, to transponder costs), and that means you need at least 1,000 subscribers.
Average data rate of 60 Mbps / 1000 subscribers is 60 kbps. Yes, Internet use is bursty, but this is a worst-case scenario.
I suspect most of these systems probably have 10,000 subscribers for a 10:1 oversubscription on a typical 60 kbps end-user capability.
Note that I've ignored any return satellite bandwidth.
The Ka band satellites do have the ability to have smaller spot beams, and they may use DVB-S2 variable bitrate to ramp up on cells that have no bad weather, but I don't think the SpaceWay sats are actually being used in this fashion right now.
The verdict I've heard from everyone using satellite internet: better than dial-up (but with more latency for games), but that's about it.
Satellite internet in Northern Canada (Score:3, Informative)
Most of these posts concern the Fair Access Policy limits and latency but you are probably excessively familiar with it already. I'm discussing how to solve your problem with access.
The key phrase that catches my eye in your question is that you are in remote, Northern Canada. The beams you are able to receive are at such an oblique angle you should feel lucky to get any service. As you know the satellites are over the equator. The latitude at which you are located is likely so far north that the satellite signal has to travel through so many hundreds of miles of murky atmosphere before it has to travel another 22,500 miles to the satellite. You might consider yourself fortunate that it even works at all at such a high latitude.
Northern Russia has an even bigger problem and they solved it with highly-elliptical orbits, sometimes called "tundra" orbits, but that requires some expensive ground equipment to track the birds and it's not even 100% available. The Antarctic uses huge C-band dishes and they're not even available at all times of data due to even more atmosphere attenuating the signal.
While you do not specifically mention the provider or the satellite format you are probably using a Ku-band system like Starband or HughesNet. WildBlue, while actively marketing to Canada, has most of its spotbeams aimed at just north of the Canadian border and they're at really oblique angles at that. Since WildBlue uses Ka-band it's out of the question for these distances and will be unavailable when the weather is rainy.
To solve this problem you'll be looking at expensive, non-consumer solutions that work in the C-band. Though the signals were somewhat weaker in the past, the newer satellites serving up north have surprisingly powerful C-band beams and, being in the C-band, they aren't affected by raindrops like Ku-band and Ka-band are, so the low angle in your location wouldn't be such a huge problem.
The following outfits provide this kind of specialized internet service. I hope they are useful to you.
For this kind of money, and the way C-band works, you can find dedicated transponder segments (and even entire transponders) so you will have a dedicated link.
This firm provides custom satellite solutions:
http://www.bcsatellite.net/ [bcsatellite.net]
These sites have VSAT terminals for C-band (they do exist in case you were wondering):
http://www.satcomresources.com/VsatTransceivers.jsp [satcomresources.com]
http://www.anacominc.com/prod_xc.html [anacominc.com]
Finally, you can punt and use Inmarsat terminals. They're not optimal but they can give you data in a pinch.
Re:Advice from an "expert". (Score:3, Informative)
DVB-S2 hasn't been around all that long, though. Most of the satellite ISPs are using QPSK 3/4 or 7/8s modulation. That'll squeeze you about 45mbps out of a 36Mhz transponder.
And true bulk time is measured dollars per megahertz per month.
TheSync...now there's a blast from the past from my Cidera days.
Re:Have you tried a Riverbed device? (Score:3, Informative)
In a nutshell, we have 6MB from our upstream via fiber or copper (not sure which) and we provide 6MB Internet service to our service area (500 miles away) via satellite. After hooking up the Riverbed, we had no perceptible difference between accelerated/unaccelerated traffic. After doing some testing, we figured out why: the far side Riverbed would send the request to the near side Riverbed. The near side would request the web page (or whatever) from the Net, then would calculate some kind of hash or checksum, which it would send back to the far side. Meanwhile, the far side unit would check for a cached page, and if it found a cached page, it would calculate a checksum as well. If the checksums matched, the near side would serve the cached page; if not, it would download and cache the new version.
Unfortunately, if your upstream bandwidth is the same as your service bandwidth, you get no acceleration, since the near side unit can't download the requested data any faster than the far side.
In fact, the protocol acceleration that the Riverbed devices provide isn't quite as good as the protocol acceleration that the Mentat boxes provided, so the Sky-X Mentats actually performed slightly better for us than the Riverbeds did, even though the Mentats don't do any caching.
Satellite (Score:1, Informative)
I have used satellite and VOIP extensively in the oilfield for remote locations. Spacenet offers a good service that is not over subscribed. It is certainly not DSL speeds or cost, but we have successfully run 2 phone lines, fax, and internet over the same connection with minimal issues.
Re:Get some of those BUDs in that other thread (Score:3, Informative)
Ah, good times.... Pilots servicing that air strip call their passengers 'Pang Pukers'. I'll never forget walking into an Iqaluit bar and hearing the end of a conversation between two bush pilots: "So I turned around and said to her, 'Lady, do you mind not screaming so loud? I'm trying to land the plane here.'"
I've landed about three times in Pangnirtung, and everything they say is true. But it's worth the trip. It's one of most forbiddingly beautiful places in the world.
Back on topic: Solar is right out in places like Pang. It's not worth the effort of bringing the equipment in. Wind, on the other hand....
It's all in the return channel collisions... (Score:1, Informative)
It has to do with the way satellite works, and in particular, collisions with other subscribers on your return VoIP traffic. In most cases, you're running over a TDMA (time devision multiple access) or RA (RAndom Access) return channel. This means you burst your return traffic to the bird with no clue who else is talking at the moment. Think ethernet hub, but with 700ms latency. You can hold more subscribers in this manner per satellite with less overhead. If you get a collision, it re-transmits, but this service level is entirely inadequarte for voice as you need guranteed return traffic. Latency is not an issue, as long as packet loss and jitter are ok. Latency only gives you annoying delay, which you must deal with.
Now, to get voice to work you need some provider who will offer some type of call admission control who will dedicate a time-slot to your return channel sufficient for the bandwdith of the call. Then it will work great. Unfortuantely, even 64kbps of dedicated return channel spectrum is expensive. Your only hopes to find this economically is some system which flags, classifies, and provides dedicated bandwidth on deamnd for the duration of a call, and then releases it to other subscribers when you hang up. I mean really, dedicated bandwidth specturm (khz), which then is bandwidth is expensive for a megabit, expect to pay $10,000/mo (now I'm sure they pay less in quantity and have fewer capital costs). For a 50kbit VoIP dedicate return channel that's around $400/mo. Just think of the oversubscription level necessary to sell that to you @ $100/mo. With normal intrnet usage only, 80 people on a T-1 sized line is entirely possible. Or VoIP usage if it only holds that channel upon during the duration of the call is quite economical too.
This is all assuming the outbound (hub to you) is not just oversubscribed, but that should never have loss outside of oversubscription. The return will always have loss, even if it's not oversubscribed under TDMA/RA, just less loss.