The Wintel Treadmill As Seen By Gartner

Gartner reflects on the long time taken by businesses to migrate away from XP. They recommend three alternative strategies to avoid these problems with “7”.

Strategy My take…
“Deploy Windows 8 on new PCs as they arrive, thereby phasing Windows 7 out over time as PCs are replaced — this may make sense for many organizations.” This assumes a treadmill model of PC-deployment, a constant stream of new ones replacing old ones. Why? There is no business case to replace anything in business periodically if it’s still working, not chairs, not tables and not PCs. The longevity of XP was partly due to the longevity of the PCs bearing that OS, nearly 8 years. If the OS breaks sooner, change it, not the PC.
“Skip Windows 8 and plan to deploy a future version of Windows (perhaps Windows Threshold or even a release after that) to replace Windows 7 — we believe most organizations will do this. With this strategy, many will not eliminate Windows 7 before support ends unless they budget extra funding to do so.” This is exactly what businesses did with XP. Where’s the recommendation to avoid XP-itis?
“Deploy Windows 8 on all PCs to eliminate Windows 7 — for most organizations, we see little value in doing this, and do not recommend it without a solid business case.” Exactly! This also means there’s no value in replacing “7” with any future version. Conversely, one can replace XP or “7” with GNU/Linux and be better off forever: less malware, fewer re-re-reboots, no Patch Tuesdays, no stream of cash for licensing, forever, etc.

No. The correct solution is to just get off the Wintel treadmill. That makes every move in IT make business-sense. Bolstering M$’s business at the expense of your own makes no sense.

See Plan Now to Avoid Windows XP Deja Vu With Windows 7.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology and tagged , , , , , . Bookmark the permalink.

32 Responses to The Wintel Treadmill As Seen By Gartner

  1. DrLoser says:

    The answer in Australia is 4 years for a General PC with Windows or OS X installed(yes if you dig into the Australian god darn tax code it states the operating systems). Laptop is 3 due high breakage rates(and this one does not state the operating system so this is what a tablet is yes tax code lets go nuts).

    Yes, pretty much what I said, oiaohm. Why do you feel this constant need to claim what other people say as a novel assertion on your part? It’s distinctly peculiar. Are you a leader, or a follower?

    There is no define in Australian tax code to cover a server or a thin client.

    … and, for consistency, this is immediately followed by …

    Max figure is 50 years on computer hardware in Australia.

    Which means that there is “a define.” What a surprise.

    And anybody who thinks that they can get away with presenting a 20-year amortization for a thin client in a saw-mill of all places is just plain barmy. I mean, I’ve met some dangerous nutters in my short time working in wooden structural engineering, but I’ve yet to meet anybody who seriously believes that any sort of electronic equipment is going to last twenty years in an enclosed area with all sorts of nasty flying around.

    Not to mention that the chances of being able to hook up said “thin client” to any useful application on any useful server at all in more than, at the most, eight years are vanishingly slim.

    This is getting to be enjoyably absurd, oiaohm. Give us more of your electric-fencing Ancient Greek typography Moving Objects Through The Power Of The Mind lunacy on the subject, please!

  2. oiaohm wrote, “Amortization processing is very different country to country.”

    When the tax-man is involved that’s very true. Otherwise a business could use whatever makes sense for them. In Canada, depreciation is a certain percentage annually, so the value never quite goes to zero. Thus, the tax-man gets a bite even when the thing is junk… Typically, depreciation is allowed as a tax-deduction for businesses. Of course the rate varies dramatically from one category to another. Interesting things happen if the goods, like buildings, actually increase in value…

  3. oiaohm says:

    DrLoser Australian tax code allows for instant deprecation in case of failure. USA tax code does not. So the year the device explodes and its not under warranty you get to deprecate whatever is left in Australia. So a thin-client that explodes inside 3 months and you got it second hand you could deprecate it completely in the current tax return in Australia.

    Amortization processing is very different country to country.

  4. oiaohm says:

    http://depreciationrates.manager.io/
    DrLoser 20 year thin client is what you find in sawmills for logging log data.
    Amortization is a purely accounting concept, dealing with the depreciating value (perceived, to a business) of a particular asset.
    The thin client role in sawmill is not going to change over the next 20 years. So value to business does not change for 20 years.

    There is a problem here sorry I forgot USA tax code is different to Australia. USA you have 5 years to deprecated a fork lift even if it sitting in storage doing nothing. Australia min is 11 years. So you go by operational hours in Australia so you can deprecate a well used forklift sooner.

    The Useful life idea is not a international thing.

    The answer in Australia is 4 years for a General PC with Windows or OS X installed(yes if you dig into the Australian god darn tax code it states the operating systems). Laptop is 3 due high breakage rates(and this one does not state the operating system so this is what a tablet is yes tax code lets go nuts).

    There is no define in Australian tax code to cover a server or a thin client. Max figure is 50 years on computer hardware in Australia. Amortization rate must be less than 50 years. You must have documentation to back and explain the lifespan in case of an Australian Tax Office audit being pain.

    A heavily used server/pc might be deprecated in 1-2 years. Of course if you are deprecating this fast you have to document for the tax man in Australia.

    DrLoser I will give you 20 years on a thin client in USA accountant has made mistake due to USA tax code absolutes. 20 years on a thin client in Australia not always a error. 50 years on a thin client in Australia might not be a error either.

    Tax code is nicely globally different.

  5. DrLoser says:

    To put this plainly, oiaohm, if an accountant assigns an amortizable timeframe of twenty years to a “thin client” of any description whatsoever, that accountant will either be had up on legal and even professional grounds, or will simply be fired.

    Whether or not that is a reasonable “lifetime” for a thin client (I would assert that it is not, not even remotely), it’s an utterly absurd number in accountancy terms. It requires a level of predicting the future (eg replacement technology) that is far, far beyond what amortization is primarily in place to do.

  6. DrLoser says:

    Amortisation is based on the useful life of the hardware. So if Window 7 machine did not reach zero when it went end of life there would be something wrong with the Amortization maths.

    Not really, oiaohm. Not at all, in fact.

    Amortization is a purely accounting concept, dealing with the depreciating value (perceived, to a business) of a particular asset.

    Since nobody in their right minds calculates depreciation retrospectively, what happens is that the accountant assigns a specific value to that asset — normally, in the case of hardware and of commercial software packages — at the point of purchase.

    In order to calculate the depreciation, the accountant will then make an assumption about the “useful life.”[1] About twenty years or so ago, my understanding is that this figure was typically three or four years. I believe, and am happy to be contradicted by somebody who knows, ie an accountant, that the current figure is five years.

    Once that five years is up, the asset is fully depreciated, and appears on the books at $0 cash value. It might last another twenty years (to use your farcical and easily disproven case of a “thin client.). It might have blown up after six months of use.

    Either way, the book value of an amortized asset at the end of its depreciation is precisely nothing at all.

    [1] The “useful life” in this context refers exclusively to use within the business, doing whatever it was purchased to do. This does not include paving it over with Ubuntu, or any other form of taking a dusty bit of garbage with blown capacitors and turning it into a doorstop.

  7. DrLoser wrote, “said PC would have been bought with 7 on it”.

    3 years ago, I was working at a school that obtained 12 NIB PCs with XP on them. We paved them immediately after I demonstrated to the students how slow they were compared to our 8 year old thin clients.

  8. thr wrote, “the new Mayor Dieter Reiter was the one who in his prior job as City Councilman for Labor and Economy was responsible for Microsoft building their new HQ in Munich.”

    Yeah. I suppose M$ invested $millions in his promise to become mayor and convert the council and staff to non-Free software. [extreme sarcasm]That’s a no-brainer.[/extreme sarcasm]

  9. oiaohm says:

    Amortisation is based on the useful life of the hardware. So if Window 7 machine did not reach zero when it went end of life there would be something wrong with the Amortization maths. A thin client bought at the same time as the Windows 7 machine might have a 20 year operational life so a lot slower Amortisation. Fan-less Linux machines have a slower Amortization as well.

    DrLoser you don’t understand Amortization. The fact something has been deprecated to zero means the company should find money to replace it. The longer the operational life of Linux systems can result in lower overall costs and slower Amortisation.

    In fact introducing Ubuntu or so on to a Windows 7 machine can rebirth from a deprecation point of view. So now its life becomes when the hardware fails.

    2016 Windows 7 machine in a Windows only network will be deprecated to zero. Windows 7 machine in a business that is hybrid Linux/Windows the machine may not be deprecated to zero and it will depend on what the machine is made of. These could be identical model machines. This is something people are not aware deprecation rates are not identical company to company. Forklifts for example are deprecated based on number of operational hours. Linux hardware is deprecated the same way as the forklift. X hardware has so many hours of dependable operational life.

    So its all how Amortisation is performed. If someone performs hours of functional life against most Windows networks it comes very clear that a lot of hardware is being disposed of before it dead.

  10. DrLoser says:

    DrLoser, feigning ignorance of bundling, wrote, “That ain’t “eight years” right there, is it, Robert?”
    A PC bought 3 years ago by a business running XP could well last another five years, eh?

    Robert, feigning ignorance of both standard corporate practise and amortisation, fails to notice that said PC would have been bought with 7 on it, and would have been depreciated to zero by 2016.

    At which point, Windows 7 still has four years to go.

    Bit of a silly little tactic that, wasn’t it, Robert?

  11. thr says:

    “thr mentioned M$’s new HQ in Munich. Munich is the IT capital of Germany. That move was planned years ago and has nothing to do with the current mayor.”

    Yes, it has, because the new Mayor Dieter Reiter was the one who in his prior job as City Councilman for Labor and Economy was responsible for Microsoft building their new HQ in Munich. He is also a self-confessed Microsoft fan.

    Don’t worry, Robert. The re-migration won’t take 10 years. This time Munich will hire people who know what they’re doing.

  12. thr mentioned M$’s new HQ in Munich. Munich is the IT capital of Germany. That move was planned years ago and has nothing to do with the current mayor.

  13. thr says:

    “Fortunately, the city council is democratic and the mayor and his side-kick get only two votes. Most of the council and its employees are happy with GNU/Linux. This is old news. It was published weeks ago and nothing came of it.”

    Oh, something will come of it, Robert. Especially since Microsoft Germany is relocating to the heart of Munich.

  14. DrLoser wrote, “Vienna moved to Linux, Matts?”

    It’s a work in progress. At a recent presentation, it was revealed that the last negotiation with M$ was in 2008 after a pilot of GNU/Linux in 2005/6. They have lots of FLOSS on servers and clients. Their present view:“Situation 2014 – OSS is reasonable usable (server and client)”

    Currently they use lots of FLOSS including GNU/Linux on servers. There really isn’t anything holding them back from going to GNU/Linux on clients except making the decision. They have tight budgets that are probably the last nail in the coffin of M$’s OS there.

  15. DrLoser, feigning ignorance of bundling, wrote, “That ain’t “eight years” right there, is it, Robert?”

    A PC bought 3 years ago by a business running XP could well last another five years, eh?

  16. ram wrote, “Gartner has long been a MIcrosoft shill.”

    Well, so has just about every other “consultant”, “journalist”, and “expert”. It was so easy to make a living that way for decades. The tide is turning however, M$ cannot afford to pay off everyone and M$ is no longer a growth industry on the desktop. Gartner, IDC, etc. will be looking for an exit or hedging bets. Android certainly gets lots of play with them. OEMs and others interested in small cheap computers certainly don’t want to be told M$’s is the one true way…

  17. ram says:

    Gartner has long been a MIcrosoft shill.

  18. oiaohm says:

    https://www.kernel.org/doc/Documentation/circular-buffers.txt

    DrLoser
    Now, unless you can explain to us all how a Linux device driver, or a Linux scheduler, or a Linux file system, or a Linux network stack — which incidentally would necessarily involve infinite recursion — can avoid that latency
    As normal you have asked the wrong question. The question is how has hardware evolved to allow OS latency to be avoided and why is Windows not using it.

    The answer is circular buffers + accelerators on hardware. Linux kernel circular buffers match the circular buffers implemented in network accelerator hardware and other accelerator hardware. The result is Userspace-Networkhardware-Network and reverse with no kernel space activity other than set up the connection to the device. You could refer to Linux Network stack as software emulation in case network accelerator cannot do it or you don’t have a network accelerator.

    So on high performing hardware Linux network stack fairly much does nothing.

    We are also seeing circular buffers being used by SAS controllers and this can result in the strange event that a fuse driver(userspace file system driver) is in fact faster than the kernel space driver for the same file system due to cost saving not having to remark memory between kernel space and userspace.

    Linux scheduler under Linux you have many options. But its the introduction of circular buffers for many tasks that allows a CPU to be decanted to running particular applications constantly. So fairly much a no scheduler mode. Result is application will not be context switched away when a message comes in.

    File systems are still performance hell. You cannot give all or majority of file-system operations over to a decanted chip so limiting how far hardware acceleration can go in this area.

    Linux as a OS core over time is in fact handling less and offloading more and more.

    There is a problem. Darwin(OS X kernel) and BSD kernel(Freebsd and so on) implement the same kind of circular buffers as Linux right down to using the same in memory design. Windows does not.

    That Paul E. McKenney from ibm was involved with it you can be fairly sure its patented with the condition that any OS using it has to release source code. This is would not be the only case that Windows is lacking a low level feature because it closed source so cannot use particular patents.

    Why do some things by IBM and others to use a patent require source code be released. Its to be able to make sure the implementation exactly match.

  19. oiaohm says:

    http://www.sogo.nu/ there is a difference with Linux world now. There is a tested solution to replace Exchange.

    What is Munich using “Kolab Enterprise”. The talk of lack of unified desktop client by the Majors is lack of support for old versions of Outlook. 2013 outlook will sync against most open source group-ware servers using activesync this include kolab. Prior to 2013 you require openchange todo it without MAPI plugins that Kolab does not support and the MAPI plugins for Outlook are many grades of disasters as Outlook updates break them.

    Kolab Desktop Client(kontact) is huge in size on Windows. This is the problem part of the argument was lack of unified client. Reality there is a unified client the majors just did not like it. Yes the email client argument end with Outlook 2013 as well. Yes is is why part of the dispute now that Outlook 2013 can be used against open source servers.

    This is a lock in combination of Outlook and Exchange will die over the next few cycles.

  20. DrLoser wrote, “all of this functionlity depends crucially on latency…”.

    That’s a partial truth, not even the half of it. Read M$’s EULA. They make the user of all their software depend on a dozen limitations and costs that are now going to be in-their-face rather than bundled with the price of a PC. See? Now consumers do get to shop around and pick a winner rather then allowing M$ to stack the deck.

    ie. Android/Linux clearly is the consumer’s choice even when M$ spent hundreds of millions on ads, made all kinds of sweet deals and sued the world. It all came to nothing for M$. Their share of the pie is shrinking.

  21. thr wrote, “Mr. Pogson, good news: Munich is getting serious about switching back to Windows. About time this LiMux abomination dies a quick death.”

    Fortunately, the city council is democratic and the mayor and his side-kick get only two votes. Most of the council and its employees are happy with GNU/Linux. This is old news. It was published weeks ago and nothing came of it.

    Whatever the opinions of the new mayors, the smooth operation of the present system and the horrible cost of switching yet again are not something most welcome. They certainly did not campaign on this issue and the council and citizens will not support them on it.

  22. thr says:

    Mr. Pogson, good news: Munich is getting serious about switching back to Windows. About time this LiMux abomination dies a quick death.

    http://sz.de/1.2090611

  23. DrLoser says:

    Ok, let’s revise my statement. No one will want to have a WINDOWS OS that is based in the cloud.

    That’s a remarkably percipient observation from somebody like you, Dougie, who is a self-confessed educational failure. Somehow or other, you have actually managed to get to the nub of the matter. Want to know why people won’t want this?

    Because an OS deals with device drivers, schedulers, file systems, and these days network stacks. That’s basically what an OS does. It’s all it does.

    All of this functionality depends crucially on latency..

    Now, unless you can explain to us all how a Linux device driver, or a Linux scheduler, or a Linux file system, or a Linux network stack — which incidentally would necessarily involve infinite recursion — can avoid that latency …

    You’re just blowing smoke up your backside, aren’t you?

  24. DrLoser says:

    Oh, and while we’re at this, let’s bash on the difference between “Mainstream support” and “Extended support,” shall we?

    Naughty old M$.

    XP was out of “Mainstream support” in 2009 and limped along in “Extended support” until 2014.

    Somehow, almost nobody noticed the difference. For five years!. So, let’s check out the dates for Windows 7, shall we?

    End of Mainstream support: 2015.
    End of Extended support: 2020.

    It’s looking like you all will have to wait quite a few years before the Rapture, isn’t it?

  25. DrLoser says:

    The longevity of XP was partly due to the longevity of the PCs bearing that OS, nearly 8 years.

    Tut tut, Robert. Such impoverished simple arithmetic.

    Try subtracting 2001 from 2014. Allow for August in the first case and April in the second.

    Trust me. Amongst many other bits of knowledge seemingly absent on this site, I am well-versed in simple arithmetic.

    That ain’t “eight years” right there, is it, Robert?

  26. DrLoser says:

    Vienna moved to Linux, Matts?

    That’s even news to the estimable Glyn Moody.

    Never mind. Say it three times, and it will become Truth.

  27. Mats Hagglund says:

    Munich, Wien, Torino and several other cities, towns or provinces have already moved to Linux and will save a lot of money every year from now on. Millions of students are already using ChromeOS and other Linux-based systems.

  28. dougman says:

    Ok, let’s revise my statement.

    No one will want to have a WINDOWS OS that is based in the cloud.

  29. Markus Glanzer wrote, “I don’t think that there will be a Desktop OS from Microsoft after the next one.”

    M$ is so huge and the installed base so huge that I think it will take a lot longer than that to kill off their client OS. They could, for instance, give it away for $0 or even pay OEMs to install it for a decade or more with their current pile of cash. Now that Gates and Ballmer are gone, though, I don’t think M$ will do anything crazy like that. Pushing $0 licences for low-end clients is just about building the installed base, not the new plan. They will enjoy slaves in business working for free for decades still most likely. That can happen because the cost of licences to M$ is just a small part of the cost of running most businesses, not a big item. Servery is different though. Guys who have thousands of servers will give GNU/Linux a lot of play. Guys who have just a few servers love the flexibility/freedom/performance of GNU/Linux and M$ has no great pull. It’s the client stuff with all the lock-in.

  30. Markus Glanzer says:

    Well, Linux really is up to the task in most areas, IT Businesses will need to adapt, whatever the market dictates.
    I don’t think that there will be a Desktop OS from Microsoft after the next one.

  31. dougman wrote, “No one will want to have a cloud-centric OS.”

    That’s exactly what ChromeOS is supposed to be. Many schools want that because their business is not IT. They want to educate students, not work for M$ creating slaves. Many businesses are not in IT. They want to produce goods and services for consumers, governments and other organizations, not slave for M$. The cloud is a viable concept but it’s not quite mature yet. In a few years, the idea of everyone having their own server racks will seem quaint. Networks are orders of magnitude more reliable these days and even if folks grow big and have their own data-centres, they still depend on a network. I think encrypting data locally for storage on the cloud makes sense today. Soon processing that data in the cloud will make sense too. The web is evolving. We are at about web 3.0 today. Very soon the cloud will make sense for anyone with any complexity/size to their IT. Individuals with smartphones can already live in the cloud. They just don’t need their own data-centres.

  32. dougman says:

    Problem is no one wants Windows 8….LOL.

    Seriously, does anyone think that the next version of Windows will be any better? No one will want to have a cloud-centric OS.

Leave a Reply