Europe's broadband bird goes up tonight The Hylas-1 telecoms satellite is fired into space tonight, with the mission to provide satellite broadband everywhere in Europe. If all goes well, broadband not-spots could be a thing of the past - for residents who can afford to pay the relatively small premium for the service. The Hylas-1 is "mated" to an Ariane 5 rocket, …
Mobile isn't reliable. It may not connect or only give dialup speeds.
It's much worse than dialup for reliability and can easily be slower than ISDN.
Here is why even 21Mbps Mobile is inferior to 1Mbps ADSL.
http://www.techtir.ie/comms/mobile-vs-fixed
Also mobile data is currently subsidised as much as 100:1 by voice. 3 sells Mobile data aggressively in the hope you will use their voice service. At least If I was a shareholder I'd hope that's the reason they do it.
The lottery terminals most definitely *do* use satellite.
It was installed in lottery retailers premises around the same time as the 'new' lottery terminals about 2-3 years ago I think and it's been an ongoing upgrade as ISDN has lost favour. ADSL is a second choice as it's not considered reliable enough.
They certainly do. Round about 25,000 of them, with a further 2,500 on ADSL for those places that satellite isn't possible, like listed buildings etc.
3G/GPRS turned out to be a bit too unreliable, which would have put the cost up due to extra antenna etc. Satellite worked out the cheapest and most reliable, although a little more bandwidth for the price would have been a nice thing.
As I understand it, Italian firm GTECH, which provides many lottery machines worldwide, was the first to use satellite networks for its lottery terminals. I know, for example, that Virginia's lottery went satellite when GTECH updated them to new Altura terminals about three years ago. I've also read reference to satellites being in use in other lotteries such as Idaho and California.
A guaranteed unbreakable stable connection from everywhere to anywhere via Space Satellite Communications to Mother Ship IntelAIgent Control Senders and Servers, dDelivers Immediate, De Facto Virtual Machine Make Over for Leading Systems Take Overs. ........ or if you Prefer, Virtual Machine Take Over for Leading Systems Make Overs.
Sunshine to Proliferate Titan Rain Clouds of Eastern Boundless Beauty and Treasured Grace is an Immaculate Tool and Magical Mystery Turing Key, whose Possession with Exercise is a Crazy Task in the Perfect Quest that Virtual GODs Perform ...... Doing their Global Operating Device thing.
Communications paints the World Picture ..... Communicate a Beta Picture, Paint a Better World.
Ka-Sat, not Hylas1 is the big Not-Spot Bird. Launch in Dec
http://www.techtir.ie/blog/watty/ka-sat-countdown
Viasat + Eutelsat Partnership. Serious resources. 80+ Ka Spots and 70Gbps traffic.
However, Hylas1 is Avanti's 1st satellite. Up till now they have been a reseller.
http://www.techtir.ie/blog/watty/hylas-launch
Mostly ESA funded as it's partly an experimental platform. Otherwise Avanti would still be just a reseller. Only 8 Ka-Spots of unknown capacity.
My country of residence has light touch InterNet censoring, mainly to keep rabid anti-government propaganda and Facebook, away from the population.
My company installed HongKong satellite-based InterNet feeds to our offices which means we have no politically inspired 'service interruptions'. Hopefully the WTF band of do gooders that the UK government has appointed to keep InterNet users minds pristine and free from natural instincts can't get their hands on his service.
How much more of the Blunkett/Blair/Brown legacy has yet to be removed?
"Satellite broadband still suffers from a second or two of latency, long enough for the signal to get to geostationary orbit and back."
Let's do the math. A satellite in Clarke orbit is 36,000 km up, so the ground-to-space-to-ground distance is 72,000 km. Radio waves travel 300,000 km per second. That gives a 240 millisecond delay due to transit time. Yet you're right about there being one to two seconds of latency in a satellite broadband connection. I see it in the HughesNet connection I'm using now.
Bottom line: the problem is NOT the transmission time to and from the satellite. It's not an inevitable result of the laws of physics, as most people assume. The great majority of the delay is happening earthside, probably in the network operations center. It's an annoyance that the network operators could improve... if they cared to make the effort.
If latency would be 250mS due to the distance and normal lag for a datacentre is 10mS then that's 260mS. Not that good but if stable then hardly a huge issue, skype will work.
There have been many sat systems and they all have a 1000 to 2000mS latency. It simply can't be down to a poor land line to the uplink dish. It must be something fundamental with implementing a laggy IP link to satellite electronics or something.
This 250ms is simply the travelling time. Add few ms at the satellite and the ground station. Lets say that brings it up to 500ms (I've no idea, it's just an example). Well, that's an extra 500ms just to _reach_ the internet. This is not getting you any data. It's simply 500ms from your PC to the internet. Whereas on ground based systems you are probably talking about 10-20ms. OK, that's the less than 2s but this is just for illustration...
"http://www.techtir.ie/comms/mobile-vs-fixed"
I have to disagree with this, other than the crippling overage charges -- it depends on the provider. Here in the US, Verizon's 3G is actually quite reliable. Speeds stay pretty stable at home. They can vary when I'm on the road, but that's compared to DSL where it's tied to my house. And, of course, for a lottery terminal, the traffic generated is not that high.
I wonder if the lottery guys got a nice rate on the satellite, since the traffic level would be relatively low -- I'm sure the terminal probably generates <1KB when a ticket is purchased, so I'd think a satellite link (where the usage is known to be that low) could cost less than having a barely-utilized dedicated copper line running to the machine.
"Get used to the term FAP. Don't get into a contract if you find out if they implement FAP. You'll regret paying a premium price for dial up speeds."
FAP, for those who don't read the link, is basically a bucket throttle -- bucket fills at xKB/sec, you get your full speed until the bucket empties. Hughesnet's bucket is relatively small, 200MB, then you do get basically dialup speeds until you get some bucket filled up. I don't like the idea, but prefer it to getting unexpected overage charges like most of the wireless plans have.
You know what, you're right. The traffic is relatively low. I specified and designed the lottery system, and it was based on low bandwidth, low cost ideas. We wanted to be able to have burst, which we get with the Hughes system, and it works. And it keeps costs down. Several years after designing the system, I still stand by my decisions and the technology. It's sound, and so is the reliability.
GTech have used the tech for years, and they played a part in specifying it. They know how to use it to the best advantage, and that's a very good thing. They were happy when we decided to use it, and so were the National Lottery Commission. I still think of my role there as a good one, although many would probably disagree :D
PS. GTech were originally from Rhode Island, and were bought out. Very good tech guys, a lot of whom still work for GTech.
> Presumably the 66cm satellite dish isn't capable of transmitting
> up to the satellite.
Presumably the relatively low latency of a dialup connection makes it a better choice for the low-bandwidth upstream requests.
I can imagine a system where smallish HTML pages are returned via the dialup connection too, so the browser can send out the secondary requests quickly, and only the larger secondary content (images etc) is sent via the satellite.
But the quoted 1-2 seconds suggests that perhaps they aren't doing that.
But lottery systems aren't really that time-sensitive (except as drawing cutoff approaches, but that's always been "caveat emptor"). If latency isn't a big issue, then it's simpler to just use the satellite both ways. It's not exactly as if even a heavy playslip stream overloads the line (each play is a handful of bytes, plus likely encryption overhead, and each response is a small bit containing ticket number, final numbers, cost, and verification).
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
