Thin Media Clients and Home Servers

Ravi posted an article on on the topic of “thin media clients vs. thick media clients“. He’s got the right general idea but fails to talk about the other end of the equation: The thing that is providing the content to these clients. The server.

He says

thin client is used to gain access to content locked away on your computers or even the Internet via the home network.

First, I want to mention a new development in these “thin clients“. Today’s “Digital Media Receivers“ (DMR, which is the industy term for this class of device) typically have relatively hefty processors and software sets on them. For example the HP DMR displays it’s UI on the the TV. It has roughly the equivalent of WMP or WinAmp on it to do this. The new development I’m talking about is making these “DMRs“ even “thinner“ and “dumber“ by removing ALL application code form the client and having that code run on the “home server“…the UI is remoted to the device via a UI remoting technology. The first (and most important IMHO) example of this are the new Windows Media Center Extenders that will be shipping later this year (and were demonstrated at CES in January). These devices use a version of the Remote Desktop Protocol (RDP) that incorporates audio and video streaming. The UI you see is the great Windows Media Center UI.

The beauty of this approach is that over time the cost of these devices will come down even further because the amount of silicon needed to support them is greatly reduced. The core code can even be added to an existing device like the XBox (there will be a Media Extender kit for XBox…see this link). And users get a more consistent user experience.

Now back to my original point: What will serve up the content that these devices utilize? PCs! In the short term these will mostly be desktop PCs (even if they are in a quiet case and made to look like a stereo component). But in the long term we’ll see other form factors that are more dedicated to the task. Because they are PC’s (running Windows of course) they will be able to host all sorts of cool application (for exmaple the Premise Home Control software I love to rant and rave about). These things already exist, but only in niches and small quantities. But soon, I predict, the “home server” will become a hot topic…


  1. Rabi Satter says:

    Your right, I totally did not talk about servers. As things start to shake out it will be interesting to see if thin clients with servers/Internet providers will win, fat clients or some combination.

    A strike against thin clients is the dismal failure of smart displays. These devices are very cool and turn the monitor into the computer. After all when you talk at your computer do you talk to the box or the monitor ;). If you are like most people it is the monitor. Yet smart clients have not sold well. Nor I think will such technology until the price comes down below a sub $500 pc. Heck I doubt until they come down to 300 or less will they become even common place in the digital homes much less mainstream.

    But here is an interesting problem. Can Microsoft (MS) make money on these devices?

    In the consumer electronics (CE) I am not sure MS can win. The problem is the cost. If I am a CE company and I have a choice between MS or Linux, economics dictate that Linux will win. Why? Money, if MS charges even the ridiculous price of $5 a device at these low price points that is a significant cost. I am of course assuming that support, manufacturing and research are the same between Linux and Windows i.e. the cost to put either one is the same. Thus the only difference is that one costs me $5 out of my price. At a $250 price point that is 2% of the price.

    So, MS has to spend money to convince CE companies or their customers that it’s software is worth the additional $5. Add on top of that marketing bill the cost of R&D, support, etc. and I am not sure that Microsoft make the economics work. Sure you can say that they make money because they will sell PCs to support those devices but is that true. I think we have started to see saturation in the marketplaces around the world i.e. a machine is simply being sold to replace an existing computer and not really adding to the pool of computers. That’s why progressively we see fewer versions of the OS being sold at its release. True eventually they over take the old OS market but that is due to it being sold on new computers.

    Linux on the other hand has no direct R&D, support or really any cost. Yes there is a people cost for development but I have to have those people whether I use Linux or MS. So the CE company can make an additional 2% by cutting out MS. And when a good company has profits in the 10-20% range 2% is an additional 10% to the bottom line.

    By the way, I am not saying Linux or MS is a good or bad choice from a technology point of view. Both have their merits and problems. What I am talking about is good old economics. A couple of good case studies are the BetaMax vs. VHS and Apple vs. MS. The winners were the organizations with the economic model that made sense to the manufacturers and vendors who sell the product and spin offs which drive more sells of the original product. By the way Sony and Apple were not out marketed. They lost due to critical mass. Beta was only sold by Sony just as Apple is only sold by Apple. Their competitors on the other hand used licensing to grow faster thus achieving critical mass faster. By the time Apple figured out licensing it was too late and they could only really cannibalize their own sales.

    So the trillion dollar question is will MS or Linux get to critical mass first?

    p.s. cross posted to

  2. http:// says:

    Your basic premise on Linux flawed. Linux is not free to an ODM (Original Device Manufacturer). The R&D costs to take a Linux distro and customize it for a device are much, much higher than you assume, and much, much higher than the cost of implementing a profile such as SmartDisplay for Windows CE. For an ODM who is primarily interested in selling HW at volume (e.g. high volume/low margins), R&D and support costs are a critical factor. By going with Windows CE these costs are greatly reduced (at least for device profiles directly supported by CE).

    In addition, the dissapointment of SmartDisplay’s is entirely about cost, as you point out. But the OS licensing costs are insignificant relative to what really caused SmartDisplay’s to be too expensive: LCD touch panels.

    Regarding Media Center Extenders…stand alone versions will be relatively expensive initally. That’s just a fact of life for a new type of device. But as they get integrated into other devices (e.g. LCD/Plasma TVs and DVD players) the cost of the components is amortized signifantly across the rest of the device (esp. with LCD/Plasma displays).

  3. Rabi Satter says:

    I have heard that comment before about Linux that R&D cost is much more. How do you come by this assumption or information? My actually assumption is that the R&D cost for either Windows or Linux based devices are the same. This is based on the labor costs for developers being the same. True in the business world a VB developer is cheap but we aren’t talking about business apps. So a developer who can design and write these kinds of applications will tend to be more expensive than your typical programmer.

    Or is MS footing part of the bill by providing developers? Which in that case you are reallying talking about shifting the cost to MS who has to add it into the pricing of licensing or eat the cost. Which to me makes no sense to eat the cost. For the sakee of argument lets say the market is 1 billion units or at $5 billion dollars total market for MS. This means for MS to make a return on their investment they can not invest more than 3 or 4 billion total. Otherwise they might as well invest the 3 or 4 billion in other areas. They would make more.

    "insignificant relative to what really caused SmartDisplay’s to be too expensive: LCD touch panels. "

    I got to question that because for example Dell will see you a computer with an LCD panel for $499. If LCD is the significant cost how can Dell essentially give away an LCD panel. I am assuming that for SmartDisplay’s we are not talking about tiny screens. Yet a 15" Airpanel will cost you almost $1k. Ok to be fair lets price a laptop which the cheapest at Dell is $799. So why is Airpanel more. If anything it should be less after all the SmartDisplay OS should be cheaper than XP home. And the craddle can’t account for the $200 difference.

    By the way, personnal I hope Windows wins because then I don’t chuck 10+ years of Windows OS skills out the window 😉 I just question that economics and market forces automatically assume that MS will win.

  4. http:// says:

    My data comes from the fact that I work regularly with ODMs building devices. They are starting to realize the costs of engineers salaries needed to take a Linux distro and customize it (and support their customizations) are much higher than they expected. Now, remember I’m talking about device profiles that are tier 1 for Windows CE; those where MS has implemented the solution end-to-end (like SmartDisplay, PocketPC, SmartPhone, Portable Media Center, Media Center Extenders …). In these cases the ODM can almost literally use the CE platform builder to generate their image, maybe do a little driver work, and maybe a little UI branding. They are assured of very high quality levels because it has already been rigourously tested by MS. For devices where MS has not invested in the vertical stack for that profile the advantages are not as great.

    Here’s an example I’m very, very familiar with (for the previous 1.5 years I worked in this group): Windows Powered Storage (NAS). Ok, so the analogy isn’t perfect because (1) we’re talking about the NT/W2K3 codebase not CE, and (2) the price point is much higher, but the results are the same. Windows Storage Server now commands over 50% market share for NAS. Prior to us being serious about he business 2 years ago we had a miniscule share with the majority being Linux, NetApp, and EMC. Alomst all the share we’ve picked up has been at the expense of Linux! Why? OEMs and ODMs can get Linux for free? Why would they burden their products with the price of Windows Server? (Windows Powered Storage is completely based on W2K3). Because they can be assured of a high-quality, 99% complete solution that requires very little software development on their part. These guys aren’t software companies…they are hardware companies.

    Regarding the SmartDisplay topic: The $499 DELL you mention is a promotion to draw people in. It is heavily discounted in order to draw people in…in reality very few people buy the base model (Dell works very hard to up-sell you and have you accessorize). 128MB/40GB and a 2.4GHz processor. All massively commoditized components (probably $120-150 BOM). And the 15" LCD that comes with it is also now a massive commodity ($100 BOM). This all supports my point. But what really makes my point is the fact that he LCD on the SmartDisplay is a VERY different beast: it is a touch screen! Add battery, craddle, and other things…and the fact that most of it is not mfg. in commodity volumes. To be clear: the reality is this sort of device is expensive; not because of the Windows license, but because of limited volumes. Time will fix that and hopefully MS will be able to have the highest value software stack available with the lowest overall cost. For us enthusiasts, we just have to continue to be patient…

    I know I’m biased, I do work at the Big House after all. But I do have real experience here. I was one of the earliest proponents of using RDP as the basis for ultra-light-weight client devices in the home w/in MS. SmartDisplay and Media Center Extenders were both conceived by me. I’m just fairly tired of people continuing to say "Linux is free" without really thinking through the total costs. The good news (for MS anyway) is we continue to see strong evidence that various companies are waking up to this.

  5. Rabi Satter says:

    This is an enlightening conversation. From my perspective it will help me understand the market from a business perspective. And to have facts to backup my assumptions.

    As far as the NAS analogy it actually is a good example. True the OS from a cost perspective is more. However in terms of percentages it should not be that off base. So from this perspective you are really contending that R&D is more for a Linux than a Windows based product because MS provides a turnkey environment. On the other hand a Linux environment requires someone paid by the manufacturer to create the software. When you cost it out against the licensing of Windows it comes out for the sake of argument to be the same. So now at this point you look at the support and quality of the platform. In the case of a Linux platform, the manufacturer has to fix bugs, create upgrades, etc. In the case of MS that cost is borne by MS. So if your core business is hardware manufacturing the value proposition of using MS is so you don’t need to also be a software developer.

    Or put another way, MS saw that providing a reference platform for manufacturers is the value added proposition that justifies the licensing fee. Where as Linux proponents, no one seems to have realized that the value proposition of creating a Linux company that focuses on creating reference platforms and making money by licensing the platform.

    Is this a fair assessment?

    In regards to the SmartDisplays, I still do not understand the pricing model. As you point out the hardware is not volume hardware. And that is why I am surprised! If anything it should be volume hardware configuration with the only added cost the touchscreen film. After all to run RDP should not require heavy duty horsepower.

    Having been on a touchscreen design team it did not cost that much to design or run a prototype batch. The biggest hurdle was sound and screen resolution since we wanted a small high res screen. Which was the issue. If we had said a 15" high res screen it would have been cheaper vs. a 12". Go figure. The sound was a side issue since we wanted to have a fairly nice stereo sound coming out of the screens.

    Last thing to be clear. I would not ever say that Linux is free in terms of development. What I meant by free is this; that if R&D is equal (which you point out is not true) then the price of a Linux device vs. Windows device would be the difference in the price of the Windows licensing fee.

  6. BT says:

    I have thought about this a lot and even written my own software to do it (PlayerPal). At the moment I think the perfect control hardware is the Pocket PC. They are very cheap and support Wifi and HTTP.

    As for remoting the UI, I use good old ASP.NET and .remoting to remote my servers WMP. Thus my server is headless (no keyboard, mouse or monitor).

    I think a lot of the current DMR UIs suck. IR control is also a bit arcane in the Wifi age. I have designed the UI to be Google like. It is optimised for surfing/discovering your music.

    Anyway, if you are interested the PlayerPal remote control for Windows Media Player can be downloaded over at It’s not the holy grail but I hopefully it will contribute in some way.

  7. Tom Swart says:

    There is some thought of switching from PC to Thin Client on our campus. I have talked to some larger campuses and they say not to do it. That PC is still better? Whats is the straight story on this? Is the PC becoming a dead issue and thin client the real deal? I’d like to be better informed on this topic. Any comments would be appreciated.

  8. Rabi Satter says:

    Depends on the application. In this case we were discussion a particular type of thin client.. i.e. media clients. These devices are great for that kind of operation because of environmental issues… i.e. noise in the living room. Keep the device cheap, silent and that way your stereo, TV, or Theater does not have to overcome the noise of the computer.

    I think that thin clients are useful under other kinds of conditions. Specifically, ones in which you do not want storage on the machine. For example, using a thin client machines in a public setting for Internet access or in customer service centers really most company jobs. Here the machine is just a PC with no drives of any kind or USB ports. It uses bootp to boot the OS from a server and only can use applications that are available from the server. This makes support much easier since the machine can only be in a configuration specified without any programs being put on by unauthorized personnal. The user can still store stuff like documents file servers.

    The biggest problem is cultural since people think of their office PC as their’s even though it is owned by the company. So people feel they have the "right" to put additional software on the machine. Hence I think the adoption of the so called thin client has not been faster.

    By the way, in writing this I can’t really think of a situatiton where an organization would get a lot of benefit from deploying a network with thin clients i.e. storageless PCs and large fileserver farms. There are even companies that specialize in making thin client farms i.e. a rack that holds a huge number of diskless PCs with keyboard/mouse and monitor connected by cat5.

Debate this topic with me:

This site uses Akismet to reduce spam. Learn how your comment data is processed.