Ildana to get Nobel Prize

jwt

Following this PR piece http://www.siliconrepublic.com/news/news.nv?storyid=single5994 and in particular

It also overcomes the inherent round-trip time of other satellite services, often referred to as latency

Iladana are sure to get a Nobel Peace Prize for the groundbreaking work in Physics.

Breaking the speed of light barrier to reduce round trip times is possibly the most profound chage in modern physics since Einstein.

We await further details with bated breath.

John

damien

I hear their new equipment is totally environmentally friendly too as the core material is taken from water vapour.

jwt

really?

Hmm according to the latest specs posted on http://www.ravinglunatics.com it's supposed to use a cunning blend of pixie dust and zero point energy which has the added benefit of producing more power than it consumes thus reducing the enviromental impact of coal and peat burning power stations.


John

well at least I backed up my claims with some links, mind you I wonder if ravingluanatic is a real website? hmmm I can buy it

paulm17781

jwt wrote:

Following this PR piece http://www.siliconrepublic.com/news/news.nv?storyid=single5994 and in particular Iladana are sure to get a Nobel Peace Prize for the groundbreaking work in Physics.

Peace prize for phyisics? They are amazing indeed.


I read that article earlier and sighed. :rolleyes:

damien

Jesus just called. Man is he pissed. Apparently he doesn't like it when someone can out-do him. I think he's suggesting a breakdancing throwdown to see who is more powerful. That answers the "What Would Jesus Do?" question anyway.

watty

Indeed total bilgewater.

from Silicon Republic wrote:

Ildana’s Murphy acknowledged that up until now satellite broadband had been given a bad name due to latency issues but mostly due to the unpredictability of satellite coverage. “The distance between a satellite dish and a satellite is about 40,000km. Coverage generally doesn’t vary but what does vary is the mechanism companies use to communicate with the satellite. It’s mostly a case of who shouts loudest,” Murphy explained.

He continued: “It’s like the difference between a well ordered meeting where everybody gets a chance to speak and a student’s union meeting where no one gets a word in edgeways. Whether Ireland likes it or not, in terms of satellite coverage we are the Outer Hebrides of Europe and most satellites are designed for the most populated areas of Europe. Ireland gets the crumbs from the UK. Most commercial offerings don’t work this far west.

“As long as you have a predictable channel, whereby the system gets heard every 120 milliseconds, you can build bomb-proof VPN access. We built a ‘link budget’ calculation system to ensure the best network capability that ensures broadband connectivity rates of 32Mbps download and 8Mbps upload. Voice over IP quality is no different to that of a mobile phone call,” Murphy explained, adding that the company’s research and development efforts were funded through the support of the European Space Agency and the South West Regional Authority.

A "ping" has to go 45K kilometers up, down, bounce via internet then up down
If we take 40ms For internet
ignore latency in any buffers, receivers, transmitters (which is actually significant).
distance = 45 x4 = 180million km
Round trip approx 480ms + 40ms, or about 1/2 a second.

In practice VOIP needs codec buffers that add latency and the TX/RX add latency too as the TCP/IP is converted to a different protocol that handshakes the packets at both ends assuming the satellite trip will work, or else round trip time of non-UDP traffic would be 960ms.

Another trick is caching, though with more non-static webpage interactive content this can "break" web pages as those using caching proxy servers have found.


I think IBB use a Satellite to connect the Clarion in Limerick as normal Skype round trip times are less than 400ms but to IBB user they are 1790ms minimum, terrible delay on a conversation (I'm sure they don't but why is it so bad?)


You can have 200Mbps on Satellite if you like, but it won't fix the latency.

Satellite is a method of last resort when even ISDN or dialup is not available. Amazon Jungle, Himalayas, Sahara etc.

Even Antartica has an optic fibre Backhaul...

jwt

paulm17781 wrote:

Peace prize for phyisics? They are amazing indeed.

Yep , they are going to scoop both at once, physics for the astounding strides forwards in advanced physics and the Peace Prize for bringing to an end the ongoing struggles for broadband in Ireland otherwise known as "The BB Troubles"

John

paulm17781

The sad thing is the general public will buy this!

I had a web designer tell me about the 100mb wireless connection with satellite back haul he is getting in his new house through a GBS. I was too drunk to explain latency to him so I just agreed. He is relatively technical and hadn't though about RTT, what hope does Joe public have of understanding this?

watty

Does the General public have enough money to afford it though?

paulm17781

watty wrote:

Does the General public have enough money to afford it though?

I meany buy as in believe.

viking

I recall the Mullingar GBS seminar a while back where I gave a short talk on the pros and cons of each technology that communities might be considering for GBS. Straight after I finished and sat down, the Ildana rep stood up and spoke the same tripe thats in the SiliconRep article. I questioned that I'd be interested to see how they managed to overcome the latency associated with a 45Km an back round trip to a satellite, he said he had some equipment set up in another room which he intended to show the public and would happily demonstrate it to me after the seminar.

After the seminar, the demonstration I saw was a Powerpoint presentation on a projector.... nothing else. Some demo...

In fairness to Ildana, they were the only ISP who bothered their arse to show up.

Viking

Sponge Bob

Is it a sekrit photon acceleramatator wonders this Sponge ??? Silicon republic should be ashamed of themselves , this is as ludicrous as Beam ...back in the day

damien

I miss Genesis Europe

SeaSide

Sponge Bob wrote:

Is it a sekrit photon acceleramatator wonders this Sponge ??? Silicon republic should be ashamed of themselves , this is as ludicrous as Beam ...back in the day

Nahhhh

Its RFC0800BITEME encapsulation of a flux capicitor in the IP packet header

lucernarian

I wonder do they have access to some uber-secret technology that we, mere mortals, do not know about yet??
Oh wait, horses**te is not particularly new or technological:rolleyes:

JohnC.

SeaSide wrote:

Nahhhh

Its RFC0800BITEME encapsulation of a flux capicitor in the IP packet header

Only catch is that the flux capacitor needs 1.21 gigawatts and needs to be travelling at 88 MPH.

Urban Weigl

I, for one, welcome our new light-speed-breaking overlords!

SeaSide

Kahless wrote:

Only catch is that the flux capacitor needs 1.21 gigawatts and needs to be travelling at 88 MPH.

I think thats actually gigawhats and if you drop it out of the fifth floor window it will get to that speed

uncle_sam_ie

viking wrote:

I recall the Mullingar GBS seminar a while back where I gave a short talk on the pros and cons of each technology that communities might be considering for GBS. Straight after I finished and sat down, the Ildana rep stood up and spoke the same tripe thats in the SiliconRep article. I questioned that I'd be interested to see how they managed to overcome the latency associated with a 45Km an back round trip to a satellite, he said he had some equipment set up in another room which he intended to show the public and would happily demonstrate it to me after the seminar.

After the seminar, the demonstration I saw was a Powerpoint presentation on a projector.... nothing else. Some demo...

In fairness to Ildana, they were the only ISP who bothered their arse to show up.

Viking

Was he like the guy in that Simpsons episode where he was trying to sell the monorail?


Seriously though damien.m, this is a con and something has to be said about it. There are desperate people out their that will buy into this crap. I'm just glad their not setting up in my area.

viking

uncle_sam_ie wrote:

Was he like the guy in that Simpsons episode where he was trying to sell the monorail?

Hehe, JWT had an interesting "exchange of words" with him after JWT called satellite "rubbish" in front of Minister Demspey at a DCMNR public event

Did you ever get Ildana's Xmas card after that John?

crawler

Sponge Bob wrote:

Is it a sekrit photon acceleramatator wonders this Sponge ??? Silicon republic should be ashamed of themselves , this is as ludicrous as Beam ...back in the day

No but you are close - in trying to recreate the events of the bigbang (http://lhc-machine-outreach.web.cern.ch/lhc-machine-outreach/) it was discovered that the rules of physics could be "tweeked" - makes me happy that satellite no longer has latency

johnmu

Hi All,

First of all, many thanks for the Nobel Prize nomination, and if I can reduce the latency to zero I will have earned it.

The phrase "No Latency" may have come about as a result of my saying "Low latency" when I was explaining how VPNs and VoIP can work over a satellite link. In particular, I was describing the inherent difficulties in putting together a reliable VPN and the even greater difficulty involved in getting VoIP to work reliably without dropping out.

Most of the participants on this list are quite technical so please indulge me for a moment and I will try to explain what we have done without giving away the family silver. There are essentially two types of security required for any kind of a secure transaction; security for data in transit so that it cannot be intercepted and bilateral authentication so that Billy knows he is talking to Jack and Jack knows he is talking to Billy. In order to kick-off the process, Billy needs to convince Jack that he is who he claims to be and vice versa. Once this process is complete, they need to setup a secure channel. In a nutshell this is it. In order to maintain good security, credentials may be challenged at irregular intervals and this is one of the areas where problems can arise with an inherently high-latency link. If this latency varies it can mean that it may be possible to setup the initial link but it may drop out during one of the renegotiations. This is extremely irritating for the user to the point that it will probably be unusable. It will mean that it takes a long time to establish the connection and it could drop out several times during the course of the day, and each time, setting it up again could take a long time.

So what is latency? Latency is the catch-all phrase which has come to mean the time between an application layer request and the application layer response - not unreasonably. Over a lot of satellite links this time can vary considerably from several hundred milliseconds to some seconds causing the application layer request to time-out. The real problem with VPNs is not the catch-all phrase "latency", it is this unpredictability of the latency; the amount of variation around a particular mean which can cause your VPN to fall over.

All we have done is we have put a lot of work into nailing down this unpredictability so that it is possible to build a reliable VPN across it. We have had a lot of success with this with customers achieving very high levels of througput.

In order to have a good link, it is essential to run a calculation called a "link budget" which determines the absolute minimum level of satellite strength required in order to guarantee no degradation in your network performance to a particular level of confidence taking into account rain-fade zones, seasonal activity etc. This is the backbone of the service, inadequate signal strength will mean the VPN drops out. This effect is less noticeable if the user is using satellite just for general browsing and email.

At this stage, you can or should have, depending on the transport mechanism, have a reliable link. Now the throughput should only be at the mercy of the round-trip-time or RTT which will not vary. It takes approximately 113ms for a request to get to the satellite and approximately the same to come back to Earth, and then then the same again in reverse. This is your starting point, the absolute minimum for the RTT will be approximately 452milliseconds. The only remaining problem is the throughput. Since TCP is an emotionally insecure protocol, it sends information in "chunks" and will not send the next chunk until it has received an acknowledge that the first chunk has been received correctly. This is relatively easy to do on TLS/SSL based schemes but is considerably harder on IPSec based schemes and that really is the family silver so I will not be discussing that just now.

Finally, VoIP is not unlike VPNs in that it is important to keep the variance of the channel under control. In this regard, satellite as a medium is even easier to deal with than wireless as the channel is, in many ways more predictable. There is a net delay on VoIP of about half a second, but then the delay on mobile phones can peak at more than that. I do indeed remember the conversation I had with the person in Enfield and I offered him a demonstration at our offices, not there and then. Satellite is just another medium and any technology has its issues, very little is insurmountable.

Thanks for your patience.

John Murphy
Technical Director
Ildana / Pure Broadband

johnmu

Hi All,

First of all, many thanks for the Nobel Prize nomination, and if I can reduce the latency to zero I will have earned it.

The phrase "No Latency" may have come about as a result of my saying "Low latency" when I was explaining how VPNs and VoIP can work over a satellite link. In particular, I was describing the inherent difficulties in putting together a reliable VPN and the even greater difficulty involved in getting VoIP to work reliably without dropping out.

Most of the participants on this list are quite technical so please indulge me for a moment and I will try to explain what we have done without giving away the family silver. There are essentially two types of security required for any kind of a secure transaction; security for data in transit so that it cannot be intercepted and bilateral authentication so that Billy knows he is talking to Jack and Jack knows he is talking to Billy. In order to kick-off the process, Billy needs to convince Jack that he is who he claims to be and vice versa. Once this process is complete, they need to setup a secure channel. In a nutshell this is it. In order to maintain good security, credentials may be challenged at irregular intervals and this is one of the areas where problems can arise with an inherently high-latency link. If this latency varies it can mean that it may be possible to setup the initial link but it may drop out during one of the renegotiations. This is extremely irritating for the user to the point that it will probably be unusable. It will mean that it takes a long time to establish the connection and it could drop out several times during the course of the day, and each time, setting it up again could take a long time.

So what is latency? Latency is the catch-all phrase which has come to mean the time between an application layer request and the application layer response - not unreasonably. Over a lot of satellite links this time can vary considerably from several hundred milliseconds to some seconds causing the application layer request to time-out. The real problem with VPNs is not the catch-all phrase "latency", it is this unpredictability of the latency; the amount of variation around a particular mean which can cause your VPN to fall over.

All we have done is we have put a lot of work into nailing down this unpredictability so that it is possible to build a reliable VPN across it. We have had a lot of success with this with customers achieving very high levels of througput.

In order to have a good link, it is essential to run a calculation called a "link budget" which determines the absolute minimum level of satellite strength required in order to guarantee no degradation in your network performance to a particular level of confidence taking into account rain-fade zones, seasonal activity etc. This is the backbone of the service, inadequate signal strength will mean the VPN drops out. This effect is less noticeable if the user is using satellite just for general browsing and email.

At this stage, you can or should have, depending on the transport mechanism, have a reliable link. Now the throughput should only be at the mercy of the round-trip-time or RTT which will not vary. It takes approximately 113ms for a request to get to the satellite and approximately the same to come back to Earth, and then then the same again in reverse. This is your starting point, the absolute minimum for the RTT will be approximately 452milliseconds. The only remaining problem is the throughput. Since TCP is an emotionally insecure protocol, it sends information in "chunks" and will not send the next chunk until it has received an acknowledge that the first chunk has been received correctly. This is relatively easy to do on TLS/SSL based schemes but is considerably harder on IPSec based schemes and that really is the family silver so I will not be discussing that just now.

Finally, VoIP is not unlike VPNs in that it is important to keep the variance of the channel under control. In this regard, satellite as a medium is even easier to deal with than wireless as the channel is, in many ways more predictable. There is a net delay on VoIP of about half a second, but then the delay on mobile phones can peak at more than that. I do indeed remember the conversation I had with the person in Enfield and I offered him a demonstration at our offices, not there and then. Satellite is just another medium and any technology has its issues, very little is insurmountable.

Thanks for your patience.

John Murphy
Technical Director
Ildana / Pure Broadband

damien

johnmu wrote:

It takes approximately 113ms for a request to get to the satellite and approximately the same to come back to Earth, and then then the same again in reverse. This is your starting point, the absolute minimum for the RTT will be approximately 452milliseconds.

What about the time required for processing by the equipment at all the various stages, the dish on my roof, the bird in the sky, the dish at the downlink facility etc?

This is an interesting solution

Stark

What's this? We have broadband using quantum entanglement now? Sweeettttt.

jmcc

johnmu wrote:

The phrase "No Latency" may have come about as a result of my saying "Low latency" when I was explaining how VPNs and VoIP can work over a satellite link. In particular, I was describing the inherent difficulties in putting together a reliable VPN and the even greater difficulty involved in getting VoIP to work reliably without dropping out.

You should be more careful when talking to technology journalists. While they may claim to cover technology, most of them have no technological background whatsoever. Unless the facts are explained on a drool proof press release, they will inevitably get it wrong.

Most of the participants on this list are quite technical so please indulge me for a moment and I will try to explain what we have done without giving away the family silver.

At a guess, it partially relies on varying the expected response time with a multiple channel solution. Though luckily I've more important things to do now than evaluating satellite system security.

Regards...jmcc

watty

It isn't too complicated. A number of published solutions (all involve converting to and from TCP/IP at each end of the link) in simple terms acknowlege the packets before they go to satellite, like as if each end of the satellite link has a proxy server and the connection between the proxies is NOT TCP/IP...

You still can't get latency approx. less than 1/2 second. It is still a technology best suited to broadcast.

Satellite should NOT be needed in a developed country like Ireland. It is a technology of last resort for Internet, for very isolated mountanious regions, large deserts, great forests and vwery poor places with no infrastructure.

It would take little money in therms of overall Irish ICT spending and Infrastruucture to give everyone in Ireland either adsl, Cable TV BB or reliable Wireless Broadband.

All the Satellite providers I've talked to convert the vanilla TCP/IP to avoid handshakes over the Satellite path.

None are very happy about the user trying to host anything.

Contention is either terrible if the service is reasonable priced and successfull, or if contention is good, then the service is exhorbitantly expensive. Satellite Two way is the most expensive bandwidth in the world.

If there was sensible line priceing, you could bundle multiple analog lines (four = 200K) and get 50ms RTT with a much cheaper non-contended 200K approx both ways data.

bminish

watty wrote:

Contention is either terrible if the service is reasonable priced and successfull, or if contention is good, then the service is exhorbitantly expensive. Satellite Two way is the most expensive bandwidth in the world.

Watty,
The contention issue is mostly down to transponder costs, Do you happen to have a ballpark figure on what renting a transponder would cost (per Mb) for decent (I.e 80Cm or less dish size) would cost?
Presumably up-link transponder bandwidth suitable for dealing with up-links from 80Cm sized Dishes is in relatively short supply and even more expensive than down-link bandwidth. These costs preclude anything approaching reasonable contention at reasonable cost

I have been on Satellite, I have done quite a bit of digging into the proxying tricks. Browsing can be made satisfactory over VSAT providing you are not dealing with Secure sites where the end to end encryption makes the proxying tricks ineffective. No mater what stunts are used doing anything interactive over a system with a 500 msec+ delay is going to be painful at best.

There is no place for Satellite Internet in Ireland, Nowhere in Ireland is remote enough to justify it and it's just a poor excuse for a lack of joined up thinking.

Can someone who is on an Ildana system/group scheme supply some ping and trace routes to a representative Irish site such as www.heanet.ie. My bet is for the minimum RTT time to be no less than 670msec and the average will be much higher than that.

Ping times from InshTurk to www.heanet.ie are about 11 mSec. (3 radio hops to Galway then on to ESB fibre.) If this can done for the Offshore islands off the Mayo coast why do places only a few Km from fibre need Satellite solutions?

brendan.

watty

One transponder only gives a shared bandwidth of about 25M bps. You need two transponders, or 1/2 the bandwidth as it is duplex.

so for 10,000 users online at once on a single transponder it is 12kb/s per user.

If only 10% are online then it is 120kb/s per user.

Sponge Bob

watty wrote:

One transponder only gives a shared bandwidth of about 25M bps. You need two transponders, or 1/2 the bandwidth as it is duplex.

so for 10,000 users online at once on a single transponder it is 12kb/s per user.

If only 10% are online then it is 120kb/s per user.

I was under the imprssion that a full transponder on most modern sats like is min 33mhz to 40Mhz but am not sure how the overhead eats into this to calculate effective thruput.

eg this one is going up this week or next .

http://www.skyrocket.de/space/doc_sdat/hotbird-7a.htm

25Mbits of data shared betweeen all transponder users sounds pretty correct

A full transponder a few years back was up to £5 Million per annnum . see

http://www.museum.tv/archives/etv/E/htmlE/europeancomm/europeancomm.htm

had that fallen to as low as €1m per transponder per annum we would be looking at a cost of €40k per uncontended mbit per annum

A full transponder translates into maybe 10 tv and 10 radio channels if used for TV and Radio Broadcast ..which is what sats are great at with fewer channels if used for HD TV of course.

watty

FEC overhead, house keeping over head etc. In fact 25Mbps may be generous. Nearly 1/3rd of some DTH TV is error correction.

A larger dish allows for lower FEC, hence TV feeds (users have very large dish) may be only 7/8 FEC compared with Sky 2/3 FEC on 27.5Msps transponders.

Indeed Satellite makes adsl, Wimax, Metro and cable TV look silly on BROADCAST, with a potential for 2000 channels from one satellite to cover 200 Million viewers with same signal...

But for indivdual data streams you need to limit to about 50 users simultanously on line approx 2M bit. If this is 20:1 contention then one transponder supports 1000 users. Even at about €800,000 Euro COST price, that works out at €800 per annum. But Cost price could be much more.

The satellite operators may have higher costs than a CableTV or xDSL supplier too.

Note that according to BT, Wireless, (which is MUCH more expensive than xDSL and MUCH cheaper than Satellite), has a pay back time of 6months for ALL captial expenditure. After this running cost is tiny compared with Satellite. For Metro or Wimax wireless most of the running cost is backhaul bandwidth, which the satellite operator as to pay about as much per user for that too.

It is easy to see why decent Satellite two way BB is about 300Euro per month for about 1/2 aDSL speed, or about twice ISDN speed. (If they arn't charging this amount for as little g'teed speed, then they will go bust)

Contention is worse at any given ratio on Satellite than on Cable or adsl, because only heavy users / users almost always online are going to order such an expensive service in the first place, whereas on adsl, Wireless or cable, unless you are in an insane student village, typically you may get very few of the shared users actually using the system at the same time.