Designers hope low-scale desktop package shakes up Microsoft-dominated software market The idea is straightforward: Instead of giving employees computers packed with features they rarely use, companies could save tons of cash by distributing simple machines tied to powerful central servers. Computing vendors have had marginal success over the years with variations of this “thin client” concept. Now IBM Corp. is betting that with some tweaks, the technology can become a big hit, challenging the traditional approach pushed by Microsoft Corp.
From a management perspective, the thin client approach is a win-win in almost all cases, even for home users with multiple PC’s around the house. The more computers an organisation has, the more trouble you can save by using simple workstations, managing software installs, files, access controls and backups in a few central locations.
But: why aren’t thin clients cheap (hardware I mean, not Total Cost of Ownership) ?
If files are stored on servers, you don’t need local storage: bye bye harddrive, and all reliability issues related to it.
If you only need some common applications, you may have more options in what Operating System to use. Take advantage of that choice, and the actual hardware platform to use, becomes a variable too. Take advantage of that, and you can go with a low-power, passively cooled, low- or no-noise small form factor system.
Workstations for administrative-like purposes? So: no 3D gaming? Take something with basic 2D video support integrated in CPU or chipset (MediaGX, VIA Eden platform or similar).
So, suppose I would want something low-power, passively cooled (no moving components), small form-factor, without harddrive, CD/DVD or floppy, with relatively small memory and ‘slow’ CPU, without hardware 3D support, and audio/video integrated in the chipset: can anybody explain why available systems like that are more expensive than your average desktop PC that HAS harddrive, CD/DVD drive, 3D videocard, etc. etc.?
As long as that doesn’t change, thin clients are only a plus from the management or hardware reliability perspective, not in a hardware cost/user sense. And that is really a shame: if you can make individual systems low-power, small, and simple, then why shouldn’t that hardware be cheap as well?
Easy my friend: volume, volume, volume
This may change in the future of course.
Cheers
I was wondering when OSnews was going to cover the significance of IBM’s Workplace 2 stategy.
It is in many ways a direct assualt on MS and an attempt to preempt Longhorn. The article failed to mention a number of the key aspects of Workplace 2 that make it more than just a Lotus Notes replacement/MS Outlook alternative, that is based on Websphere:
1. It is Java based on the Eclipse platform using the new Rich Client Platform in 3.0. It develops the concept of a Rich Client – more than a thin client and less bloat than with the normal fat client/file and print server approach from MS, giving considerable flexibility in the relationship of the client to the server system.
2. By being Java based it cross platform for Linux, Windows and Macs and like Eclipse it has a plugin architecture allowing for both Java and native plugins (like MS Office on Win) and new rich client applications based using RCPML an XML based language similar to XUL. It uses SWT for its widget set so it appears native on all platforms.
3. Workplace Documents is a database (DB2) backend for transparent workgroup based filesharing. The versions of Workplace 2 for notebooks and PDA’s have a small database built in with personal websphere software to provide for a local document database, when not connected to the corporate network that can be synced to the corporate server later. The rich client approach allows for fast interaction with the corporate network even over a low bandwidth modem connection.
$. Workplace 2 has the “IBM editors” built in, a fork of Openoffice, to provide WP, spreadsheets and presentation handling with format compatability for MS Office documents. Making it suitable for use on Linux computers and systems without MS Office installed.
It is a very subtle attack on MS in the corporate desktop arena. It is aimed as much as a preemptive strike against MS’s aim to move more into deeper control of computing in the enterprise (by means of the XAML and Win FS technologies in Longhorn) as being a scheme driectly to replace Windows by Linux on the desktop (though providing room for a foothold). How successfull they will be, I don’t know but I wish IBM every success as they are offering a cross platform standards based technology that can sit entirely on top of an open source infrastructure. I am glad the my company uses Notes as its global groupware platform as it may well be more likely to follow IBM on the desktop than MS at least I hope so.
What an original idea
0) Buy a cheap $50 router and connect a decent computer to it as well as a few old computers.
1) Enable Remote Sessions in GDM on your “server”
, which you can do through GDM’s nice GUI.
2) Download PXEs iso and stick in the DVD drive of an old computer. http://pxes.sourceforge.net
Yeap, your old pentium 133 suddenly runs KDE at kick ass speed and you can manage all clients from a single computer.
Install kiosktool and you can lock down the clients as much or as little as you desire, making it very easy to set up a baseline that is common to all desktops.
Have fun.
I think not. Thin client mode *is* in the best sense of the meaning, ‘traditional computing’. My, how soon we forget those IBM S360’s, etc…
Thin client systems are not cheap because the price of rolling out a server farm capable of supporting hundreds of user sessions is far greater (and price increases exponentially as more users are added) than simply giving everyone cheap user-side computing.
From the article:
Workplace 2 runs on Windows or Linux computers, and its dashboard can incorporate the big three applications in Microsoft’s Office software package — Word, PowerPoint presentations and Excel spreadsheets.
….A Macintosh version is due this fall.
So it’s not a thin client. The author sort of blows up it’s importance. It’s just a cross-platform collaboration framework with some pre-made apps.
I understand it better as chemicalscum put it, but it still doesn’t seem much different than enterprise Groove. Or maybe just an advanced/ extensible version of Exchange/ Outlook or Notes.
But: why aren’t thin clients cheap (hardware I mean, not Total Cost of Ownership) ?
One of two reasons:
If you’re building a thin client out of off the shelf x86 hardware, the only real difference between a regular machine and a thin client are no hard disk and less RAM. Neither of these items add a great deal of cost and hence, reduce the price much.
If you’re building a thin client out of lower volume, more “custom”, non-x86 hardware then they cost more because of lower volumes and less common technical expertise.
The only real major advantage of thin clients is manageability – unless you’re buying in phenomenal volumes, the hardware cost saving is basically nonexistant. Of course, selling manageability as a benefit to PHBs can be _extremely_ difficult, which is why thin client solutions aren’t as common as you’d think.
I have a book about computers that says that “thin clients” are going to be the main form of computing within the next 5 years.
The book was written in 1975.
“Thin client systems are not cheap because the price of rolling out a server farm capable of supporting hundreds of user sessions is far greater (and price increases exponentially as more users are added) than simply giving everyone cheap user-side computing.”
Nonsense.
We spent 2000 bucks for a server that serves 40 thin clients made of old Pentium Is.
Total cost:2000 and a day of work.
And I believe our solution would scale even further with a little more RAM. So I don’t buy this bullshit. I am in the field and what you claim doesn’t wash with what I am seeing.
Thin client systems are not cheap because the price of rolling out a server farm capable of supporting hundreds of user sessions is far greater (and price increases exponentially as more users are added) than simply giving everyone cheap user-side computing.
Say what ?
Cheap (~$4000AUD, ~$2000 USD), commodity hardware will support dozens of thin clients running “standard” tools like Office, web browsers, email and custom DB frontends with no problem whatsoever. Scaling (from a hardware performance perspective) is basically linear – got more clients ? Buy more hardware.
Thin client hardware costs – at *worst* – are no more than standalone client hardware closts and have the potential to be quite a bit less (if your rollout is sufficiently large). They certainly aren’t anything approaching “exponetially” higher.
Wyse Winterm stations range from 299.00 to over 1000.00, many others will be joining this market shortly which will drive down costs. The drop in the price of server hardware has really made this whole idea come together.
Some of us “oldtimers” remember tossing hundreds of these Wyse units in the trash 15 years ago. Now we are donating all of the desktop pc’s to charity, and dragging the dumb terminals back in.(interesting cycle)
For many companies desktop computing is a thing of the past, and for many others it soon will be. The impact this will have on the world pc market/IT field is going to be nothing shy of huge. It is certainly going to take it’s toll on Intel/AMD, as well as the hard drive makers and others.
The only real problem with thin client networks is they remove desktop support/jr.admin/local admin jobs at lightspeed.
At work when we installed our Laboratory Information Management system, IT choose a Citrix based thin client approach. The existing lab PC’s running NT4/2000/XP dependant on age (some were as slow as 200 MHz machines) were loaded with Citrix client software connecting to a a small farm of Citrix servers and the LIMS software on Win which iterfaced to a small Red Hat cluster running the Oracle back end. It worked a treat so our whole lab data system software was was ported of to the Citrix thin client system as well.
The advantages were for users it didn’t matter how old your computer was it ran as fast while the advantage for sysadmins was considerably increased ease of maintenance. For me I could run the Citrix ICA client on my Linux box and have full access to LIMS and laboratory data processing.
All round it has been a great success and a lot of old client machines don’t need to be upgraded urgently. On the server side in exchange for improved manageability it has required a modest increase in servers as we are only getting 16 clients per server, of course on a Linux based thin client system you can get quite a few more clients per server than with Citrix on Win servers.
The hardware cost will not be of any importance because to the consumer it will be free. You plug it in your broadband connection and if you paid your suscription it will work if not it will say “press OK to accept a charge of …$ on your credit card”. Like your mobile phone.
(Broadband, no wifi will be the connection).
“On the server side in exchange for improved manageability it has required a modest increase in servers as we are only getting 16 clients per server, of course on a Linux based thin client system you can get quite a few more clients per server than with Citrix on Win servers.”
Depends on what applications you need to run. Has little to do with the OS, has everything to do with the applications required.
Odd how folks seem to believe sysadmins control the business operations of corporations. They select it, and we just make the damn stuff run no matter how little sanity is involved. :>
On the server side in exchange for improved manageability it has required a modest increase in servers as we are only getting 16 clients per server, […]
What are they doing and what is the hardware (and what else does the machine do) ? Our relatively low-powered machine here (Dell PE650 – 2.6Ghz P4, 1GB RAM, IDE RAID1 disk) currently has 15 users running Office, IE, in-house Progress DB frontend (at a minimum) and performance remains quite acceptable. I can’t see why upping the RAM to 2 or 4 GB wouldn’t allow nearly linear scaling – the machine certainly isn’t CPU starved.
[…] of course on a Linux based thin client system you can get quite a few more clients per server than with Citrix on Win servers.
I sincerely doubt that (assuming you’re going to be running a comparable set of applications). It’s not like KDE/GNOME, Mozilla, OO.o et al are any lighter on system resources, and the X11 protocol is much less efficient (and less featureful) than RDP or ICA.
Odd how folks seem to believe sysadmins control the business operations of corporations. They select it, and we just make the damn stuff run no matter how little sanity is involved. :
Facts: The decision on what LIMS systems to purchase was made by a consensus of laboratory managers, directors and V.P.s with a lab background. The decision to go the Citrix thin client route was made by the North American IT dept. somewhere between the the managers involved in implementing and the CIO for NA where the bucked stopped. Simple isn’t it – I don’t see the point of the comment.
When I referred to IT, it includes the whole IT department including its director in the local company where I work, the CIO for North America and the global CIO in the corporate head office. The implication was not that our little local sysadmins were making the decisions. The way that globalization goes is that decisions that were once within the purview of our IT director are now at the global level.
Some of us “oldtimers” remember tossing hundreds of these Wyse units in the trash 15 years ago. Now we are donating all of the desktop pc’s to charity, and dragging the dumb terminals back in.(interesting cycle)
Projects like PXES ( http://pxes.sf.net ) let you recycle those pc’s saving not only reducing dramatically TCO but reducing initial investment too.