Enterprise PC Bidding III: PC Alternatives - Desktop Linux

Rob Enderle

The guts of PCs have gone through a lot of changes in the past two decades, but changes outside have been less pronounced. Even laptops, other than some minor cosmetic changes, look very similar to what we were giving employees a decade ago.


We still don't have a notebook in the mainstream as potentially useful as the old IBM ThinkPad Butterfly. Desktop computers, for the most part, are tower or mini-tower designs whose only distinctive change has been a color shift away from white to black or gray.


But there have been attempts to change the PC dramatically. The first was thin clients, closely followed by Blade PCs, and IBM almost created the modular computer, which is still floating around in concept. Intel has coined the term UMPC (ultra mobile PC). OQO was the first with one of these in the market and has as a competitor, Flybook, in Europe, but Apple's iPhone may define the success of this segment that bridges PCs and smartphones.


We'll deal with these emerging technologies in two sections: thin clients/blade PCs and smartphone/UMPC. Many, if not most, of these thin client devices run Linux and may provide the best path for that platform to the high-volume desktop market.


Thin Client/Blade PC


I'm combining blade PCs and thin clients because the two segments are starting to merge. Originally, thin client computing, as envisioned by Oracle and Sun Microsystems, was a lot of terminal-like devices connected to a large server. The idea had merit, but the technology that existed in the late '90s was simply not where it needed to be.


Arguably, the Sun Ray One was the best of the offerings in terms of performance and security, but the cost was prohibitive as was the inability to work well with an existing dominant Windows environment. It is interesting to note that one of the most common complaints from an ex-Sun employee was how horrid this system was to actually work on in terms of performance.


In a way, the concept was a recreation of the mainframe, yet mainframes' massive I/O was simply not available to servers, and the processors couldn't divide up the load to provide adequate performance. Multi-core systems are changing that, but for now, thin clients as provided by HP, Wyse and Neoware are best used where processing is minimal and data entry, running remote applications and browsing the Web are primary functions.


As soon as you need to really use the processor, and particularly if the application is graphics-intensive, thin clients stop being appropriate. This is where blade PCs come in. Pioneered by Clear Cube and now provided by HP and Hitachi as well, this a derivative of the blade server and an advancement over rack-mounted PCs and KVMs. Initially, this provided many of the benefits of a thin client in terms of a massive reduction in complexity, heat and noise on the desktop, but they were PCs, and while they used similar wire to an Ethernet link, they couldn't be switched. Switching was later added, but not without picking up the thin client hardware and running into graphics limitations again, this time having to do with network bandwidth and latency rather than server loading. The result, while not as pronounced, was still unacceptable to many.


IBM launched an initiative called Virtualized Hosted Client Infrastructure, which is the closest to a a true blend of thin clients and PC blades, as part of a big move back to the desktop after selling its PC division.


Most recently, both HP and a new company founded by ex-Transmeta employees called Teradici (currently working with Clear Cube) have found ways to provide high-level graphics capability within network limits. The result, particularly when done over fiber lines, could be the best of all worlds on the market today.


As a side note, the Teradici technology is likely one of the big ideas to hit the market this decade. It uses chips that compress and encrypt a high-quality data stream and then decompress and decrypt it on the client side. This results in a near-KVM level of thin client running only firmware, substantially increasing the security of the solution while also increasing the performance and reducing the cost significantly. This has applications that go well beyond the PC, making the company worth watching.


Blade PCs, as a result, can be used almost any place a desktop PC would be appropriate, but are particularly valuable where noise, heat, security and uptime are critical. These places include call centers, in pharmaceuticals (particularly labs), finance (particularly trading floor), K-through-12 education, government and military. Because the client is cheap and doesn't have a fan, the health care industry seems to like the solution as well, as it reduces or eliminates the need to destroy PCs should they become contaminated by a contagious disease.


The reason we aren't all on these things today is they lack standards at a card level, making it difficult to competitively bid blades; they can be expensive to operate, particularly in existing office buildings that may lack the real estate or the cooling needed for the centralized services, and they don't deal effectively with mobile workers, though this last factor is changing. Laptops now account for well over 50 percent of new company sales.


With 802.11n nearly ratified and companies such as Neoware, Itona and SafeBook fielding thin client laptops, we are simply waiting for a Teradici solution to become viable.


As we wait, however, smartphone UMPCs are starting to roll to market.


Smartphone/UMPC/Modular Computer


It is interesting to note that the OQO, the first viable UMPC that you could put in your pocket, and the iPhone, the more popular of these devices, both came from ideas hatched at Apple. While smartphones have traditionally been used for pagers and mobile e-mail, they are getting ever more capable browsers. The iPhone is arguably more powerful than some of the early Apple laptop computers, suggesting that, by version 3 when it will be enterprise-ready, it could deal with the mobile requirement left out of the solutions above. In any case, many think this class is the next big thing.


The Flybook initially came with a built-in cell phone, which buyers, most of them business executives and millionaires, used for office conference calls and to ensure data connectivity. The current generation OQO has built-in WAN, like many of the notebooks today, and could be used with even fewer data trade-offs than the iPhone. Palm recently released the Folio, which is basically a thin client laptop wedded to a cell phone wirelessly and tied with a PC blade or thin client server. That solution also is looking more attractive.


Finally, modular computers, named most exciting by IT in a 2001 survey, are back under consideration with a bunch of us old PC guys trying to refloat that idea, based on having a common core of technologies in a module that would, much like an iPod is today with music, be accessorized to become anything from an in-car data platform, a traditional-looking laptop or desktop, or something completely different and unique (like a wearable computer).


This last notion has a lot of promise, but requires the most work.


Wrapping Up


Thin client and blade PCs are relatively mature, while smartphones/UMPCs/modular computers are not yet quite ready to fully displace your PC. The OQO and Flybook, because they are computers, come closest, but are best used in conjunction with a desktop computer.


We are at the beginning of a major change though, and the vendor who can get this right first likely will define much of the next decade in the PC market. Right now, HP has a substantial lead (it has thin clients, blade PCs and smartphones), but is having problems with solutions that cross divisions, providing for a substantial amount of long-term exposure for the company.


But, if you have a major security concern or need to consolidate your PC resources and gain some of the other benefits promised by these emerging technologies, better start considering them now because it's best to be properly prepared.

Add Comment      Leave a comment on this blog post
Jul 25, 2007 6:10 AM Harold Voelkner Harold Voelkner  says:
Where is the part on Desktop-Linux, the main part in the title of this paper?? hev Reply
Jul 25, 2007 10:19 AM Rob Enderle Rob Enderle  says:
Thin client products (mention this in the opening) are arguably the most successful Linux based desktop offerings in the segment. Folks forget that so thought I would call it out right in the title. "Many, if not most, of these thin client devices run Linux and may provide the best path for that platform to the high-volume desktop market."A lot of folks follow Linux on the desktop. Reply
May 14, 2008 1:02 AM Ian Ian  says:
"Arguably, the Sun Ray One was the best of the offerings in terms of performance and security, but the cost was prohibitive as was the inability to work well with an existing dominant Windows environment. It is interesting to note that one of the most common complaints from an ex-Sun employee was how horrid this system was to actually work on in terms of performance."I'm a former Sun Employee/ Contractor , and I'm reminded of my days just last year working on a Sunray 1G......The systems were rather slow, mainly due to the implimentation of the system server side. Also, the fact that all applications were held on different servers, not to mention we were halfway across the country from the systems. There was also poor capacity planning, it wasn't that the system of thin clients was bad, it was the fact that there was poor implimentation.The performance was what you would expect based on the limiting factors. However, you can use Sun Rays on Windows Based Systems. To set it up within a windows based enviorment involves using VMWare to emulate a Windows 2003 enviroment. I know there is some documentation on it within Sun Solve. (I no longer have access sadly otherwise I would give you the document ID/ You can call and ask them they should be able to help) Reply
May 15, 2008 8:54 AM Rob Enderle Rob Enderle  says:
Agreed, the SunRay was impressive, but man was it expensive and the Sun employees I've known complained about it a lot due to its performance problems. But it did showcase a path that, had it been done better, should have been more successful. Reply

Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.