Sunday, April 24, 2005

Ericsson, Cisco & IBM aim for Latvian IP TV

A consortium consisting of Swedish telecoms giant Ericsson, internet infrastructure builder Cisco and IBM is bidding to build an IP TV network for Lattelekom, this blogger has learned. In this case, IP TV means delivering DVD quality image and sound over 100 + channels over a DSL type connection at around 20 Mbps.
The ongoing bid also explains the eagerness of IBM's representatives in Latvia to demo what it said was "its" triple-play product. To be sure, the slides had many other company names and the IBM people said they were only "integrators", which is the role assigned them in the three-member consortium. At the same time, the IBM guys did an excellent job of leaving the general feeling that somehow, it was their "show".
The fact that Lattelekom is set to build out an IP TV network is nothing new for the readers of this blog -- what is interesting is that IBM was packaging it as "triple play". That means there is a very low cost, and, in many cases, practically free internationa and domestic voice element in the solution. Indeed, I would call "triple play" more like 2.5 play, because with the heavy duty internet pipe needed for IP TV, one has the capacity for most high quality do-it-yourself VOIP solutions. So the third element is probably going to be a Cisco VOIP phone handset that makes phone-to-phone calling easier, as well as making it possible to charge extra for the equipment and, possibly, minutes on some kinds of calls.
As things look now, Lattelekom has quietly let the "soon after Easter" goal for some kind of experimental IP TV pass and is apparently looking for a serious, heavy-duty pilot level solution that will almost certainly involve the same partners it will keep for the final, commercial roll-out. I don't think it will be that soon -- maybe a Christmas offering?
If I were Lattelekom, I would hit the market with a "shock and awe" offering - voice, interactivity, games, and a hard-disk based set-top box that would allow recording of programs. I would position it so that the middle to higher income end of the draugiem.lv crowd (275 000 18 to 40 year olds who feel at home on the net) would go after it. In fact, I would make part of the launch on draugiem. This is no partiality to draugiem, just that this is one of the largest and most focused attention platforms available in Latvia after TV itself.
The IBM demo, run off a server in the basement with snippets of real programming (I think this is what they were doing) looked impressive, especially the video-on demand features and the ability to package one's own set of subscriber channels (and skip the half-dozen Russian language channels that seem to come with local cable packages, since I don't speak Russian). Interactive features are also part of the solution, so Baltcom TV, which is going digital with its own MMDS or (limited) optical fiber solution should start watching out. So far, Baltcom is the only player offering triple play, but when Lattelekom gears up and starts, Baltcom will face some very serious competition.
However, a caution to all is that despite its high number of cable TV subscribers, Baltcom has relatively few triple-play users (partly due to geographical constraints of their network).

9 comments:

banton said...

IPTV and seemingly unlimited content are all fine and dandy, but I'm afraid Lattelekom will face similar problems as with the Latvian Government initiative for an e-Government platform.

Currently the pricing for DSL connections in Latvia is absurd, this scale of IPTV would require atleast ten times faster connections - even though only domestic connections, but since IPTV requires some Quality of Service (QoS) it means that at least 10Mbps per client should be available when watching.

In the capital area this might work in high-density population areas, but the infrastructure just can't handle it with it's current set up of microwaves and haphazard fiber optics.

What I'd like to see with IPTV is low-cost content and lots of it. Which would mean that content partnerning would be essential - or what they thought that they would stream a live video from the front page of Draugiem?

I have some thoughts touching this topic in my blog.

Juris Kaža said...

Thanks, Markus.
I think the government's e-government problems are slightly different. There is a lot more real and latent demand for IP TV content than there is for, say, updating your position on the EU livestock database with your spring lambs.:)
The access problem is another story. On one level, it is having any access at all, since most people in Latvia don't have home PCs, never mind anything as cool as -- home Macs :). The other is the DSL/speed issue.
Theoretically, Lattelekom can deliver DSL in most of the country in one form or another (DSL over cdma-450 may be one future solution). Pumping 10 Mbps is another story, though I think they are gearing up for it in Riga. Right now the Apollo network (for HomeDSL subscribers) runs at 4 Mbps, I think. up from 2 Mbps earlier. The standard HomeDSL speed has been doubled to 512 Kbps from 256 Kbps with price unchanged.
The best deal is around 1 Mbps that you can get for LVL 19.95 from Telia MultiCom. Some of the deals from the "Sharashkin & Sons" (fly by night) providers might sound better.
I don't think Lattelekom will start commercial IP TV without having most of Riga and perhaps Ventspils and Daugavpils or some other places ready for it. That may be why the rumored pilot launch around Easter never happened or has not been announced.

banton said...

True, true. But the trick in pricing with fast connections is that they actually don't promise anything - Which is ok with normal internet usage. That's why a "consumer" 2Mbps connection costs x amount and a business version costs x + 200Ls - why? I mentioned Quality of Service which essentially means that the actual speed from a to b is the one promised at minimum. This means that the network company will have to exclusively reserve a slice of the bandwidth cake for the particular client. And QoS is what IPTV requires, it just doesn't work without it. Having IPTV without a dedicated IPTV QoS stream and starting up a Bittorrent download will kill the TV stream. Not nice.

Ok, so we need dedicated streams? So what? The problem is that the structure is built with 34Mbps (theoretical maximum) Microwave links which IN THEORY could hold maximum 3 users with a QoS 10MBps stream. A Single housing complex with 150 apartments would need 1,5Gbps DEDICATED connection just for the IPTV . Of course there would be redundancy issues and all connection would have to be duplicated to ensure the service.

I took this idea in my blog entry in handling by using a cache of geographically located proxy servers that have the most popular content closer to the viewers.

The only thing the trio would need to do is to build a Metro Area Network (which Cisco has invented) consisting of high-speed fiber connections and a mosaic structure layout - this would of course mean a price tag the size of conquering a country in the Middle-East. Without the oil.

If a 2Mbps decent quality connection costs 200 Ls a month, about twice the minimum salary per month - how many subscribers are they expecting?

banton said...

Well, ain't I constantly commenting here. Like a sign from above, my loyal Newsfire gave me the link of what Latvia would need and with what price.

Hong Kong Broadband Launches 1 Gbps Home Service for US$215/month

wx said...

Markus here is forgetting that TV does not mean 10Mbps link is required for EVERY and EACH subscriber - They don't need to deliver 10Mpbs x (times) number of subscribers.

TV specifics is that the same feed is pushed to everybody, so they only need to deliver the (10Mbps) stream once, and split it very late (like using a multiplexing hub once the pipe goes into a house).

and since they are already GUARANTEEING 500Kbps to EACH and EVERY DSL subscriber, I say 10Mbps stream is peanuts to deliver to a house.

Marcus, you read this right, DSL actually is all about GUARANTEED speed. You get your 60KB/sec download, that's guaranteed.

OTOH, I usually get 500KB/sec download speed from my somewhat fly by night provider, not guaranteed at all, but seems to work all the time :) And that's like only 10Ls/month

banton said...

Bah, I hate to argue, but I'll have to.

Imagine 100+ DVD-quality TV-channels (probably more, but for the sake of reason let's say 768kbps streams, the 5.1 audio takes half of that already) and the same number of radio stations of 128kbps. There's usually about 15% lost bits and/or similar inefficiency (headers, checkers etc.) meaning we need (100 x 768 + 100 x 128)*1.15 = 103 Mbps.

IF the IPTV will be build as an On-Demand solution then the network requirements will be smaller. As per my model mentioned in my blog.

Since currently the subscriber amount will be rather small and dispersed, Lattelekom has to offer the connections per household. Especially if the solution will be delivered per household.

For the quality of service part, wx, check your service providers contract and tell me the penalty for them if your connection doesn't work for a day or two? Do they guarantee that you will receive the full 512kbps per second from everywhere not regarding the lag of the server etc.? No.

Service Level Agreements and QoS which are mandatory speaking of IPTV will raise the price tag, whether you want it or no. Or do you want to watch a stream that's loading and lagging constantly compared to a normal digital television broadcast with a hard disk receiver?

And sure you get 500KB with a 2M maximum ADSL, since it's cached and your OS doesn't know how to calculate it correctly.

Do you think that Lattelekom has enough money to install a gigaether or fiber switch to each subscribers house without moving the costs to the consumer? bah, I say.

Also where the hell are you going to dig the fibers in Riga, the cobble stones are a bitch and Brivibas is the only long strait with normal asphalt and it's constantly full so you can't stuff it with traffic. Microwave antennas cost about $5k USD a piece and having three per customer makes the price tag a bit nifty for TV if LTK would push all streams at once.

And this doesn't even take redundant connection into consideration.

wx said...

just to answer a minor nit from Markus about what guaranteed speed means (as it is stated in contract for DSL, as far as I remember):

It does mean that you have your 512Kbps bandwidth allocated for you only and if you do manage to fill it up (downloading from BitTorrent, or from a speedy local server), you will have your full bandwidth always, your download should max out at 60KB/sec, not less.

Anyway, back to actual discussion,
it's pretty amazing that just a 100Mbps line (your standard Base 100 ethernet) will give you enough bandwidth for all that digital TV and even radio maybe. Well maybe two 100Mbps lines actually, because IP is quite wasteful protocol:)

banton said...

Exactly! I can't wait to see someone implement MPEG4 encoding with a bittorrent delivery method and hook it up with an working On-demand solution in a city where you can have a 100Mbps connection (like in Hong Kong) for 20 bucks a month in every house.

Of course, living in Riga might not be the best place for waiting this, but hell .. Eventually :)

I just hope LTK will do this properly, since Latvians already remember the e-Government-money-well-spent-fiasco too well to give bad rep to the IT industry and its solutions. Maybe Juris can do some sort of publicity stunt regarding the benefits of smart IT.

Juris Kaža said...

Well, I will try to give the paper the story, although I might have to source it. I was, hmm, not in Milwaukee last Friday and not meeting with some higher ups at Miller Beer, but rather seeing the big E in .se on deep background on some questions if you get my drift :). Plus I saw the IBM dog and pony show in Latvia on Thursday. Funny how they seemed to be saying they put together both the dog, the pony and the circus tent, but old habits die slowly. Also, getting this story into a certain non-white newspaper will involve getting the usual dim bulbs to light up.... enough said. Somebody from over there may actually read this.