'25,000 Windows Server 2003 boxes' must be upgraded A DAY to meet OS support death date About 11 million systems are running Windows Server 2003, according to HP estimates – some 438 days before Microsoft pulls down the shutters on support for the OS. That means more than 25,000 systems a day, starting right now, need to be upgraded to continue receiving crucial updates past the cutoff date, assuming HP's …
No doubt members of the (mainly US) product vendors cartel are delighted to have the opportunity to cash in on consequences of their respective declarations of obsolescence and the resulting knock-on effects. But can the world continue to afford to spend an increasing proportion of its IT expenditure regularly throwing working system on the scrap-heap and having to start all over?
"Luckily we didn't bite."
I'm glad we did. We have completed 3 forced major upgrades of all our Linux systems (well known enterprise type Linux distribution) that were out of mainstream or application support, in the same ten year period that our Microsoft Server 2003 boxes are still supported in. And with Linux you couldn't just upgrade for a major release - clean install only is all that is supported! Needless to say we tend to be moving away from Linux where there is a choice.
Didn't you learn in school, Capitalism is all about generating the maximum waste plan possible without regard for long term consequences.
This is called stretching the sustainability envelope, and is a specialisation of American cartels (abusive monopolies) - sorry, socially responsible, highly innovative companies.
"Claunch said IT systems running on out-of-date server software may "cease to operate correctly" due to "some latent defect that has been triggered by changes in the client's use"."
Umm, is that a threat? Since we're talking specifically about users who bought their server about ten years ago and have left it to fester in the corner, the only changes that are likely are ones coming through the update channel.
"Nice server you've got there. 'Twould be a shame if something updated it."
Makers cant support all OS 4evah. Not unreasonable to declare a cutoff date. However., the failure of so many users to update suggests that the "new" software suggests that it offers no benefits, or worse, is a pain. At what stage is good enough sufficient for the life of the hardware or organisation ? OTOH, does HP flog fog, sorry cloud services ?
If OS support is going to span 2 or more lifetimes of equipment (e.g. 10 years or more), then half-way through, Microsoft should:
1. Release a successor OS (they usually do this no more than 5 years apart).
2. Stop allowing the older OS to be sold either pre-installed or as a post-sale install.
3. Stop producing new versions of software (SQL Server, Exchange, Office related stuff etc.) for the older OS.
That way, users/orgs will naturally gravitate to the successor OS in the second half of the lifespan of the older OS, so that when the deadline looms, you won't get stupid numbers of people still on the older OS like we've seen with XP and now with Server 2003. Of course, this idea is actually far too sensible for Microsoft to actually consider. :-(
This pretty much describes the actual process Microsoft uses.
All MS products have a declared lifecycle you can check here:
https://support.microsoft.com/lifecycle/search/
This details the dates for the end of mainstream and extended support.
1. There have been 4 successor operating systems (2008, 2008R2, 2012, 2012R2)
2. Licenses for 2003 have not been available for a very long time.
3. None of the current server products (SQL, Exchange, etc) will work on server 2003.
Having known about the end of support for some time (I keep an eye on the lifecycle doc), most of the 2003 servers I am responsible for have already been decomissioned. All the rest will be gone well before the end of extended support date.
1. It gives them an excuse to get rid of the dead wood without having to go through a careful decommissioning process.
2. It keeps the paychecks rolling in.
As usual, #2 is more important than #1, but you can sell your boss on #1 because he doesn't want to pay for the clean up either.
"Install Linux, a decade worth of support with incremental updates until hardware dies or becomes completely obsolete"
That option might be available now, but historically Linux has offered much shorter lifecycles than Microsoft's OSs. Plus it costs far more to license a supported version of say Redhat or SUSE than the equivalent version of Windows Server. Then you need to allow for the cross platform migration costs. And then Linux tends to have a higher TCO for most uses other than webservers. So not really such a good choice for most companies.
"Trash hw, buy new hardware, buy new server OS license"
Generally hardware is fully depreciated after 5 years and starts to cost stupid amounts to support by then. And it will be much slower than more recent hardware, but yet still cost the same to license / support, etc. And you will likely be investing significantly in the OS upgrade / migration costs. So buying new hardware if the existing is over 5 years old is definitely the sensible choice.
Not entirely sure what point this post is trying to make. It mentions using linux to get a decade of support in an article discussing a Microsoft operating system that has had more than a decade of support.
Any discussion of OS choice on el reg always seems to ignore the main point. What software do you need to run? If the software I need to run is only available on Windows, I will use that. If linux, I will use that. If I am running up a terminal server to provide Windows based applications to end users, I don't have much choice in the matter.
OS choice is a practical consideration for me, not a philosophical one.
Regarding the hardware, since almost all servers I work with are virtual, the hardware is irrelevant. Once the hardware becomes out of date, I can just get a new host and move the existing VMs. The software refresh cycle becomes completely independent of the hardware cycle.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
