Welcome to The IP Development Network Blog
Monday, 6 August 2007
Arootz: One to Watch
Sometimes, you see something so elegant and simple, you wonder why no-one thought of it before. An old colleague and regular visitor to this blog contacted me the other day to advise me to check out Arootz - and I'm very pleased he did.
A Relative Unknown
I wish I could claim an exclusive, but I can't. They have been in BusinessWeek & the Jerusalem Post already but it appears that the point was lost on most people. There are only 53 blog entries on the company, the vast majority seeming to repeat BusinessWeek verbatim while most of the rest focusing on the fact that the company raised some cash recently.
One article I found does add some value, commenting on the BusinessWeek piece rather than just repeating it - more questions than answers, concluded Businessofvideo.com - and I agree with that part at least, but perhaps Mr Rayburn had a lot on that day because he didn't actually ask the questions. I have asked the questions - I contacted Arootz and their CEO replied with a significant amount of detail - and having looked into it, I have to disagree with the negative thrust of what Dan says.
Old Technology, New Ideas
It is true that multicast has been around for ten years or more. I know, I was at UUNET 10 years ago when UUcast was being hyped and developed in parallel. As with most other attempts to use multicast, that product failed to find a market because in the end, all it was doing was replacing broadcast and as we all know, if something isn't broken...
Multicast has again come back into people's thinking recently as IPTV services have been rolled out using the technology for the linear (live viewing) portion of what they offer. The problem there remains that multicast has not catered for timeshift behaviour. If you want on-demand, IPTV has to unicast and that means that you use a whole stream all to yourself.
I was discussing this problem with an eminent industry architect back in April at the ISP Forum event - his suggestion was staggercast, which effectively means a multicast stream of a programme being distributed every N minutes, much like Sky's multistart service for prime time movies. It was a definite improvement on multicast / unicast combinations already in use, but doesn't really tick that on-demand box.
Businessofvideo.com is also correct in highlighting that personal storage has also been around for ever. Quite right, it has, but what Arootz has done is combine this with multicast so that the network sees one stream and yet everyone gets a copy that they can watch what they want on demand. It's a mashup of two very well understood technologies and that is the simplicity that I refer to in my opening statement.
The Solution in a Nutshell
In summary, there are 3 elements - Distribution Servers where the content owner injects content, a Multicast enabled network and a set of user Multicast-2-Storage (M2S) agents sitting on PCs or STBs. Arootz sells this CDN as a managed service to content owners and works with the ISPs to make sure that multicast is turned on over the network. I'll come back to the web of relationships later in the article, but I will focus first on the service piece.
My initial reaction was: ok, sounds good they've dealt with on demand but if you are multicasting, you miss the personalisation capability that must be at the centre of IPTV to make it a step beyond broadcast. Erm no, they've thought of that.
"The ads are delivered to storage ... based on the advertising targeted parameters, the decision which ad to show is targeted individually (based on a doubleclick server somewhere) and then the ad is inserted in real time into the video stream but since it comes from storage, it is fast, high quality and real time." Arootz's CEO Noam Bardin.
That's clever - the media and the ads are delivered separately and reassembled to create the final, personalised media file...
Navigating Uncharted Waters
What about navigation and finding what you want among the wealth of possibilities?
"We allow users to subscribe to RSS like feeds from a variety of sources ... We provide interfaces for preference engines to assist in selection of content such as 'the highest rated channel based on yesterdays actual viewership' or 'all content with the word Shark somewhere'" (Shark is of course an example of something you might be interested in.)
Hmm, I like that too. This is the elegance - mashing up social networking, RSS and an EPG into something that can cope with the huge volumes of content...
Huge Volumes of Content
Arootz estimates that the average user consumes 125GB of content per month. Obviously it depends on resolutions: it might be a fair bit less than that for standard definition TV, but if we were talking about 1080p, we could be looking at four times that figure. Is 500GB a lot of data? I think that depends on whether you are a unicast network or a hard-drive.
Terabyte drives are the basis for Arootz's business model and that starts to explain why you have not seen this model previously. Storage has always been far too expensive to make plans like this work but Arootz reckons that by 2010, you will be seeing cost effective drives offering 5 Terabytes... At this point, the limitation is back on the network.
Multicast takes care of the core network capacity issue because as with caching models I have discussed previously, each media file need only be sent once to each exchange and not once per user as with unicast delivery. This saves many thousands of identical 2Mbps plus streams and brings us back and the point where the bottleneck is again the physical speed of the local loop.
Use Case Example
While Arootz claims that its service only uses off-peak capacity, this is a configuration option that can easily be changed. The idea is that you watch live TV via the live multicast feed. If you are not watching (or are watching and have some spare network capacity), other programmes are sent down to you and stored on your machine. You pull these up on demand.
Of course you can't download the entire programme catalogue. Lets say there's 10 channels that make up your regular viewing, you can't even download everything on each of those unless you have a very very very fast connection. Choosing which programme to download (because you might want to watch it) is the job of the M2S AI agent.
The question then becomes whether the AI is good enough to make sure that the file you want is already on your hard drive when you come to watch it. Backing that up, there's the fall-back unicast option in the event that you are feeling a bit wacky today. It looks like it might hold together.
You might even find that the model allows you to escape some of the shackles of the local loop speed as it allows you to watch delayed feeds at 720p (6.4Mbps) even though your line may only just be good enough for 480p (2.5Mbps). A 2.5Mbps connection maxed out enables you to receive something like 800GB per month. For 720p content you need the AI to give you a 40% hit rate (you watch 25 hours a week, it downloads 65 hours that you "might" be interested in).
So Far, So Good
Arootz links multicast with storage and adds personalised RSS subscription with targeted advertising. Sounds good so far doesn't it? The software assets they have are clearly well thought out and fill a growing need. But what about the issues?
The weakness of the Arootz model is that it requires each link in the chain to be working in harmony. The content distributor must plant the content on the CDN, the ISP needs to multicast the channels and the PC or STB hardware requires the M2S software to take advantage of the storage and run the channel selection process.
Conflicts in the Supply Chain
These three parties are not exactly in cahoots. The interests of content and network are juxtaposed and that led to the Network Neutrality issues of 2006. Sitting uncomfortably with a foot in each camp are the hardware companies. Only Sky, with assets in each area following their acquisition of Amstrad last week, seem remotely aware of the need to align the interests of the players in the supply chain.
Although Arootz has an elegant solution to the video unicast problem, they need all elements in the chain to see it and play along to make it work. Without any one element, it breaks down. If they work together, everyone wins. So who is pulling it through so that the vision becomes a reality? As with other efforts to bridge these gaps, is it a question of chicken or egg?
"Look at it differently – if you believe (and I do) that most of video is going to be consumed off IP networks then there is a scaling problem with the current technologies. This scaling is quality, cost and technical. The main bottleneck is the network and our solution is the only cost effective way for this to happen. It may not be us but it will be multicast based. Just like internet advertising was dead in 2001, premium content would never go online in 2005 – Multicast will have to rebound because unicast cannot scale to deliver the cost/quality we expect." Noam Bardin again.
I agree with everything he says there, but it doesn't really answer the question. What is the incentive for each party to play ball? I think the answer is actually much more simple.
Cash. Cold, Hard Cash.
The commercial model is that content owners pay Arootz either as a straight arrangement or as an advertising revenue share. Simple enough so far, but in parallel, they are selling to ISPs.
"Our offer to them is 'let us accelerate your content on your network such as VOD, Internet TV and other components'. We will then wholesale the access to 3rd party content owners and provide them with a revenue share back so that they get more content distributed on their networks, more efficient distribution (less load), they get a slice of the action and thus are part of the value chain, unlike P2P where they carry the cost but are not part of the upside."
Aha! Someone is taking the bull by the horns and putting in place a way to route the money so that the networks open up and get paid for carrying content. It would be easy to think that the netcos should be happy with cost savings, and try and keep the distribution fees to themselves, but this statement above all the others shows that Arootz is pragmatic and understands that a virtuous circle needs to start somewhere. And the hardware?
"It can be embedded in software or hardware, it provides distribution and targeted advertising capabilities" and "we are not the brand"
As with every small company, there are a million things that can go wrong as larger and better funded alternatives try and achieve the same thing. That said, the technology that Arootz has and the pragmatic approach shown by their commercial model is an excellent starting point.
Channels to markets (they are in many) still threaten to derail the company as putting together the video supply chain involves dealing with some very heavy hitters. It may require the sponsorship of one of these big players to get the ball rolling, but this is a solution where without compromises, everyone wins. Worth keeping an eye on...
Labels: Arootz, catch-up TV, IPTV advertising, ISP Products, next generation networks, targeted ads, timeshift TV, video on demand
iPlayer Technology Review
iPlayer Service Review
Help me to help you!
LUI Part 6 - Television over the Internet
LUI Part 4 - The Product Set
LUI (Leeds Unbundled ISP) - Part II
Joost: further analysis of a bandwidth hog
Mutton dressed as lamb
More on the Digital Divide