There are two ways to view the Microsoft IPTV system. Microsoft’s way and the wrong way. For those of us that have dealt with Microsoft in the past, this is not an abnormal positioning.
“Ours is the only end to end IPTV solution which includes all aspects of IPTV from content ingestion to the consumer device,” is the boast of Ed Gracyzk, director of marketing and communications for Microsoft’s TV division. Some people might see that as the only closed, monopolistic technology you would never control. But that would be the wrong view.
“Telcos either use us or they can use point to point products and hope that they will work together and hope that they will scale,” said Gracyzk.
Faultline openly blanched at the word “scale” used in this way. The scaling issue has been cleverly turned around. Gracyzk has managed to suggest that it is other IPTV solutions, not Microsoft’s that have problems scaling, despite the fact that some of the other systems are already working in prime time in systems heading towards one million customers. The Microsoft interpretation is that other systems don’t scale. The other (wrong) interpretation that people might place on this situation is that it’s Microsoft IPTV, that has yet to prove it can scale.
But he believes his point. You can tell, Gracyzk really thinks that none of the competition have been tested at the levels that Microsoft is expecting its system to work at, and there is some kind of crazy logic about that.
“Do you really think that companies like SBC Communications, British Telecom or Swisscom would spend that kind of money ($400m from SBC, the others have not yet revealed their spend) if they didn’t think it would scale to millions of customers?”
Well if he asks that, then no, Faultline certainly believes those companies fully expect it to scale. But that expectation, from senior management, not the engineers, in a telco, is not, in its own right, proof that Microsoft IPTV will scale. Whereas the operational system at Chunghwa Telecom in Taiwan, has scaled.
That’s right Faultline finally got its long awaited call from Microsoft, so that Microsoft could put to rights some of the “misconceptions,” that Faultline has published. For misconceptions, read quoting to the competition.
But in the end, after an hour of probing, buried among all the posturing, and suggestions that everything that was said by others about Microsoft’s IPTV was “either totally confused or outright lies,” we managed only to scrape together a few new facts about just how the Microsoft IPTV software works, but we’ll share with you what we managed to eke out.
150 millisecond dash
“Our channel change takes 150 milliseconds,” is the claim that Gracyzk made. Which is fine, but he said he cannot go into how this is achieved except to say that it is “technology we are patenting.” A long perusal at the US patent office revealed nothing of this, but then again there are rather a lot of patents waiting to go under Microsoft’s name.
“I can confirm that it uses both multicast and unicast protocols,” said Gracyzk, but would go no further. “It’s all done in software, we do not use a hardware tuner (who does?). It is achieved through a combination of client and server software and the network operator can configure this to their particular needs, with a lot of flexibility.” Is that IGMP multicast? “I’m not sure,” says Gracyzk, “I’ll get back to you.”
Faultline has already explained how we think the channel change works, with a server sending the required I-Frame data rapidly to a set top that changes channel, and then once ready, it switches over to the multicast protocol. This burst of unicast data may also be used to fill the set top buffer at high speed. Once the I-Frame has arrived, other frame data can now be accurately interpreted. This process saves up to 500 milliseconds which might be wasted in waiting for the next I-Frame. So is there a server dedicated to channel change?
“In our IPTV we use multiple server types and a lot of what has been written about us is blatantly wrong. The suggestion that you published that Microsoft software can only support three households per server is completely wrong. We can support 100s of households per server configuration,” said Gracyzk (Not per server, but for every complete configuration of server types).
Again Faultline invited an explanation. What are the server types? Is one dedicated to channel change, one for VoD. What else? The only server type Gracyzk mentioned was an acquisition server and this is the first stopping off point once the content leaves the head end, and this server, he says, is responsible for service delivery.
On the Microsoft TV website it shows a diagram with both a broadcast acquisition server and a VoD acquisition server, so there are in fact two such devices in the networks. There are also separately a VoD server and a Broadcast server, although how these differ from the acquisition versions of the same name is difficult to tell. In fact with just a web diagram and a refusal to go into the detail, the only way that anyone can understand Microsoft technology is to listen to competitors, who have listened, in turn, to customers that have heard the full pitch.
In the diagram there is also a notification server, and a server which links to the billing and OSS system; there’s a subscriber and system store, an application server, presumably for integration with other new applications, and a server that acts as a client gateway. Do all of these have to be servers running Microsoft operating systems.
“Well, yes because that’s what the software is written for. Our servers talk to other servers, which can be Linux or Unix, for instance, for billing and customer management, but our software doesn’t run on anything else.”
Spot the server
There is, of course, a server where encryption is introduced for the Windows Media DRM, which issues and manages keys. Though we’re not sure which one, perhaps the client gateway, because this has a little wall in front of it in the web diagram.
“One of the benefits of our system, and a key reason for telcos choosing our products is that our DRM gives us such a huge advantage. This is because it extends beyond the TV and the set top to home networks, tablet PCs, portable video players and even the mobile phone.
“None of these operators are looking at this as a stand alone business. For the first time the TV speaks the same language as other devices.” Once again this idea that Gracyzk has said something that means one thing to him and another to the rest of the world begins to emerge. He seems to think that a world where Telcos either define or provide PCs and PDAs is both likely and acceptable.
And also previously all TV content could be easily copied to external devices, unless a Pay TV system had a conditional access system that stopped it.
Most CA systems have stopped short of interfering with the recording of programs on DVD recorders to view later, although sometimes the CA stops recording of pay per view. So suddenly, in the new telco world, consumers may have to buy devices that have Microsoft DRM clients in them or they won’t be able to do what they always have been able to do, copy programming. So in this way telcos, and ultimately content businesses, will choose customer devices.
Gracyzk points this out. It’s not Microsoft that will make that decision, but the content owner and the operator between them.
But the more that Microsoft DRM clients proliferate, the less likely it is that Microsoft will feel the need to interoperate with other DRM systems. And if there is no interoperability, then another Microsoft monopoly will emerge almost over night. And with that monopoly, there will be all the legal trouble that Microsoft has had before. The humble iPod will not be able to get access to any music downloaded from Microsoft IPTV and presumably the same will be true of any future video iPod or anything like it, such as the humble DVD Recorder or DVR player. They will all pay homage to Microsoft or have their options limited.
Now all of this may be exactly what the content owner has in mind, the end of free personal use copying, or at least a shift to copying under the control of a DRM. But we’ve been contacted by 1,000s of consumers that go to a lot of trouble to buy systems that do not restrict their personal use rights, and any system that did is likely to become rapidly boycotted.
Of course Apple might always license Microsoft’s Windows Media DRM, but that would rather defeat its existing strategy, which is equally proprietary. When we put it to Gracyzk that the DRM licensing market could become a license business worth several $1bn, and that to Microsoft, this was in fact the main aim of the game, he simply said, “No, our aim is to make money out of selling the IPTV platform, and our DRM license terms are on our web site,” which didn’t entirely contradict the idea.
Microsoft DRM licensing is very low, and there is almost certainly a major issue in it even issuing DRM licenses. It has acknowledged, by its legally enforced payment to Intertrust, that it does not own all the patents within its DRM, so it cannot effectively license them to a third party, not unless someone pays Intertrust.
But even if DRM licensing from Microsoft is low priced now, we could point out that it used to charge very little for its operating system, but that’s managed to inflate over the years, so why not DRM license fees? And besides CE manufacturers would have to pay for the codec license, the DRM license and the media format, if they happen to develop on anything that is not a Microsoft operating environment.
So are we all going to see Microsoft on every other TV screen? After all, a TV service that says “brought to you by Microsoft technology,” is about the best advertising that Microsoft can get, and potentially it lasts forever. “In some case the operator wants to rely on the Microsoft brand and actively wants it on the screen. In other systems the consumer may not even be aware that it is Microsoft software inside the device. But even when we are shown, we are always the “ingredient” brand, never the main brand,” explains Gracyzk.
But Gracyzk certainly cleared up a number of issues for us. Microsoft TV Foundation software is still very much in use at US cable operator Comcast, and it has rolled it out in Washington State (around Microsoft?) and in November it will go to Seattle (around Microsoft?). Comcast bought 5 million licenses of the set top software Gracyzk points out, “They didn’t buy ‘up to’ five million copies, they’re all paid for”
This is very different software. “That’s because the underlying network is so different. It is a QAM based broadcast network.” But surely it is trying to do the same thing, albeit on a very different network “The Comcast system looks very different because it has a limited infrastructure. Our system will run on existing cable set tops, and they already have that infrastructure in place. We have been able to bring them new walled garden services, new applications and games.
“But you cannot do the same on a cable network as you can on a pure IP network. If you have VoIP and IPTV on an IP network you can have a notification across the entire system. If there is a phone call you can have a caller ID event which integrates the phone call onto the TV screen.
“The cable operators are trying to do this, and there is a Foundation TV project to do this, but it is very much harder. The phone and the TV are two incompatible network infrastructures,” says Gracyzk. Given that this type of translation and transcoding is usually the job of a set of network gateways and that these can be co-ordinated by systems like IMS (IP Multimedia Subsystem) we asked if Microsoft used this.
Gracyzk described IMS as “backoffice middleware,” and says that Microsoft has its own version in Connected Services Framework, which we understood not to be anything like video ready. “No, it’s not, but we can absolutely integrate the applications that it does cover, onto IPTV.”
IMS was defined to allow the alien world of 3G mobile to interact with the IP protocol. We can see every reason why cable operators would need something similar to keep their IP elements and non-IP elements talking to each other.
Giving it some OMP
Another issue that Gracyzk cleared up for us was the future for Alcatel’s Open Media Platform. “OMP has existing customers and we are working on a migration path for them.”
“They have a relatively small number of installations. And the old OMP staff are doing systems integration and new applications for existing customers,” Gracyzk said.
But how can there be a migration path if the new system needs new set tops, new multicast protocols, new middleware, and specialized Microsoft servers and can only use Windows Media DRM? It’s more a case of “ripping it out and throwing it away,” kind of migration.
And why would you have a migration path if the number of customers are “so small?”
“Sorry I don’t get your point,” stonewalled Gracyzk, when we raised this.
So onto the $64,000 question. Is Swisscom late with its Microsoft IPTV system because of delays in Microsoft’s software? Did Telstra throw it out because it didn’t think it would work? And can Gracyzk make sense of the weird set of hybrid trials at Telecom Italia, which use both OMP and Microsoft IPTV?
“Swisscom has just changed its approach. It was going for a “big bang” approach and now it is going in two phases, ramping up in 2006. The delays were due to a set top box that Swisscom wanted to offer with a DVR, that had an internal hard drive. The trial ran with one that used an external hard drive and they couldn’t get one with an internal drive in time. It was supplied by Thomson in the trial.”
Thomson won’t be happy to hear it’s at fault, since it managed to keep its name out of the story entirely up until now. But the explanation is fair and was a feature of the initial statement from Swisscom when it said “The technology was deemed not yet suitable for serial delivery, in particular since the set-top box has no internal hard disk and only one television channel is available.” But that statement has since been removed from the Swisscom website, so it’s hard to go back and check.
A big puzzle
“You have to realize that IPTV is a big puzzle with lots of pieces, which is why people want to buy it all from one source, the encoder, the set tops, the software, the network upgrade and the middleware. And there are always different requirements. The pieces of Swisscom’s puzzle are coming together at the end of 2005 and early 2006.
“For SBC Communications it is on target for this fall, when its first deployments will happen,” said Gracyzk.
“Telstra just decided that its network was nowhere near ready for IPTV and it was still working on content deals. And for Telecom Italia, you’ll just have to talk to them,” he said.
“In our Swisscom trial we were unicasting individual streams to different individuals within a home, and to a DVR, and only using just 1.8 Mbps of bandwidth using our codec,VC-1,” added Gracyzk, “So we know it wasn’t our software.”
What had really made Gracyzk call Faultline was the fact that in one of our pieces we repeated a Widevine claim that in one situation it had seen the security on a Microsoft IPTV system force the channel change to go out to 17 seconds.
“We offer sub-second channel change and we make a big feature of it,” said Gracyzk. “For cable and other IPTV systems it takes one to two seconds. And I’ve seen it take up to seven seconds at trade shows.”
Again there are two ways of seeing this comment.
It’s one thing for Widevine to tell us that on one occasion in a trial it took 17 seconds to change channels on a Microsoft system, due to the DRM. True or not that’s a small company attacking one piece of the puzzle and telling a story about a monopoly organization. It’s quite another when Gracyzk turns around and says that it takes seven seconds for his competitors to change a channel at an exhibition. True or not, they might see that as a false and misleading claim from an established monopoly.
But there are more details we extracted from Gracyzk. The one area of flexibility right now with Microsoft is the codec, and most operators have selected H.264 (MPEG 4 Level 10 AVC). “The network operator decides the video encoding standard, whether it’s VC-1 or MPEG 4 or MPEG 2 and we support all of them.”
“The client is written in C,” not C# as we all thought. “It is written to take as much of the heavy lifting off the set top and put it onto the servers in the network. For instance if a viewer wants to search across all types of media, whether they are broadcast or VoD, instead of running that search on the set top, with whatever data is held on the client, we run it on a server, in the network.”
“A lot of IPTV solutions suffer from this. They take a cable approach to design, whereas we take a next generation TV approach,” concluded Gracyzk.
Copyright © 2005, Faultline
Faultline is published by Rethink Research, a London-based publishing and consulting firm. This weekly newsletter is an assessment of the impact of the week's events in the world of digital media. Faultline is where media meets technology. Subscription details here.