The Life and Death of Microsoft Software

coondoggie writes "With Microsoft aiming to release Vista real soon now, they've been retiring older versions of the Windows OS. For IT outfits it's yet again time to evaluate what stays and what goes, and make plans for the future. Network World discusses the life cycle of Microsoft's software." From the article: "'Generally, it is a bad idea to run unsupported software, but there can be a business case to run it,' says Cary Shufelt, Windows infrastructure architect at Oregon State University, in Corvallis. The university still has some NT machines running in isolation in its labs. But Shufelt says there are security risks in allowing connections to legacy machines and that the university makes sure to minimize those risks. 'We don't allow [Windows] 9.x clients to connect to our Active Directory,' he says. 'But we try to stay current with technology so these issues don't typically come up.' Others say they also stay current to avoid headaches and fire drills."

Comments

All NT here

[Anonymous Coward]

The university still has some NT machines running in isolation in its labs.

All our Windows PCs run NT, from NT 4.0 to NT 5.2.

Re:Don't allow?

[jarg0n]

Yes, Yes it does

http://www.windowsitpro.com/Article/ArticleID/1521 4/15214.html/

Joke

[ch-chuck]

I *knew* if I just type "microsoft life cycle humor" into google something would come up:

The Life Cycle of Software

      1. Programmer produces code he believes is bug-free.
      2. Product is tested. 20 bugs are found.
      3. Programmer fixes 10 of the bugs and explains to the testing department that the other 10 aren't really bugs.
      4. Testing department finds that five of the fixes didn't work and discovers 15 new bugs.
      5. See 3.
      6. See 4.
      7. See 5.
      8. See 6.
      9. See 7.
    10. See 8.
    11. Due to marketing pressure and an extremely pre-mature product announcement based on over-optimistic programming schedule, the product is released.
    12. Users find 137 new bugs.
    13. Original programmer, having cashed his royalty check, is nowhere to be found.
    14. Newly-assembled programming team fixes almost all of the 137 bugs, but introduce 456 new ones.
    15. Original programmer sends underpaid testing department a postcard from Fiji. Entire testing department quits.
    16. Company is bought in a hostile takeover by competitor using profits from their latest release, which had 783 bugs.
    17. New CEO is brought in by board of directors. He hires programmer to redo program from scratch.
    18. Programmer produces code he believes is bug-free.
    19. See step 2

Re:Joke

[AuMatar]

Pure fiction. What programmer gets a royalty check?

The risk is not just direct

[also-rr]

We have a piece of Microsoft software, X. An application Y outputs its data to application X. So far so good...

It does this by (during the export process) loading the software X. Don't ask me why, I didn't write it.

Microsoft app X+1 is now available. App Y *will not export* to app X+1 because the executable has been moved and it can't talk to the new version anyway.

The App Y developers could fix this... but they wont because they have moved onto App Y+1 which we don't want to buy (not yet mature enough). App X is no longer available in the company and we cannot buy licenses for a variety of reasons (mostly due to integration and the fact that version X and X+1 running together cause major problems). There are no other export options except to pay for monkeys to retype all the data - on a weekly basis.

Software upgrades and end of support can attack you in the posterior in unexpected ways, and sticking with old software may not be an option. If you have given away the ability to make your own modifications, or put your data into formats you cannot read, you better make sure it's in your risk register.

Re:The risk is not just direct

[bmajik]

What you say is true, but i am not sure it is unique to software or even closed source software.

I visited the Mercedes museum in Germany a while back. One thing that struck me was the display of old fashioned factory equipment that was based on the then-new Otto cycle engine. The machine would have a leather drive belt that went up to a rotating drive wheel hanging from the ceiling. It seemed that there'd be one engine turning a row of linked drive wheels and each separate machine would have a leather drive belt that powered it.

I am sure that at some point, that engine broke, or a leather drive belt broke, or a machine broke. Supposing that any of the companies involved had moved on (think about the rapid pace of engine development during the earliest years of internal combustion engine deployment into factories) and would no longer offer parts or replacement units for any of the peices of this big moving puzzle.

The factory would be in a position to
- create the needed replacement parts themselves
- pay the original creator to fix the problem
- pay some new person to fix the problem
- abandon some or all of the systems and retrofit something else in its place

Now, you might say "ok, but if the engine had failed, wouldn't any engine work as long as it had a shaft outout and spun the same direction at the same speed?"

Probably, with some work. I assure you, i cannot go and put my BMW's engine in my Audi and have it all just "work". Engine swaps even when you're taking an identical engine from an identical car are non-trivial. Once you have different interfaces, lots of custom work has to be done to make things work, and it is a painful laborious process.

This would tend to suggest that retrofittability is critical in selecting the components that make your business run, which, when taken to the software analogy would suggest "demand documented open interfaces with open source software".

Yet the question arises - are any of the machines I've described still in use? Is using a leather belt still the best way to transfer power to a factory machine? Or do thinigs become obsolete not because of abandonware, but because progress has truly taken place? Now power is distributed via electricity, not leather belts and drive wheels. And the power doesn't come from a gas engine installed on site, the production of power has been outsourced to the power company. every part of this original system has become obsolete, irrespective of the simplistic, logical, obvious interfaces and boundaries.

Sometimes, it makes sense to just throw the old stuff away because the cost of evolving outweighs the cost of leaping.

And often times, the cost of compatability is high. Everyone seems to understand that one big reason Microsoft gets into security trouble is due to the desire to maintain backward compatability... the need to maintain interfaces and expected behaviors. Compatability/retrofittability/ease of integration are sometimes at odds with innovation and progress.

As an aside, if you're interacting with Microsoft Application X and require the binary, it usually means COM. Newer versions of X often include a backward compatible COM interface. Have you tried App Y with App X+1? Or are you going off of what the vendor says -- that to use X+1 you need Y+1?

Re:The risk is not just direct

[jimicus]

It runs even deeper than that.

Consider the following scenarios - all fictional, but all perfectly conceivable in any sizeable organisation:

• App X runs just fine, but is reaching the end of its supported life. Version X+1 has already been discontinued and cannot be licensed, any upgrade has to be to X+2. But there is no upgrade path from X to X+2 unless you want to re-key all several million rows of data, so you've got to go to X+1 first. However you never bought version X+1, so you don't have installation media and, as discussed above, you can't (easily) get it.
  
• App X is used exclusively by the finance department and is reaching the end of its supported life. X+1 is available, but it's very expensive. The finance director will have to sign off on any migration plan and he doesn't see the business need to upgrade - after all, version X has always worked so far. He's the one who'll be signing the cheque to buy version X+1. So what if the older version is not supported? We've not needed the support yet. In this case, technically the finance director is in the right - the change is expensive, has a risk attached and has little perceived benefit - however it might be wise for the IT department to have a plan B sitting in the wings in case application X suddenly breaks one day...
  
• App X depends heavily on Fred's Shiny Database and will not speak to anything else. The company that developed App X went out of business long ago, but their product is still critical to the business. Nobody's got around to investigating a replacement because the only people in the IT department who even knew it existed were made redundant in the last round of layoffs. Meantime, Fred's Shiny Database Company has been taken over by Ceefax Data Ltd, who are discontinuing Fred's Shiny Database in favour of their own product.
  

Re:The risk is not just direct

[hackstraw]

The factory would be in a position to
- create the needed replacement parts themselves
- pay the original creator to fix the problem
- pay some new person to fix the problem
- abandon some or all of the systems and retrofit something else in its place

Now, you might say "ok, but if the engine had failed, wouldn't any engine work as long as it had a shaft outout and spun the same direction at the same speed?"

Probably, with some work. I assure you, i cannot go and put my BMW's engine in my Audi and have it all just "work".

So, how did we answer these questions? Or at least we try?

Open standards and interchangeable parts.

Sure there is always going to be some degree of customization because the standards or interchangeable parts will never satisfy every situation and every case, but an engine swap would be impossible if it were not for standard hex nut sizes, and things like that.

Computers are still new, and they are becoming more commodity and interchangeable over time. For the most part, you can have the hardware and software of your choice and share things like the web, email, pictures, music, movies, etc. Tons of stuff.

Now, there are custom apps or environments that do not always conform to standards because there is not one, nor is there enough of a market to create one. And that is where you pick an environment, for good/bad/indifferent, block it off from the outside world, and then DON'T TOUCH IT.

Re:The risk is not just direct

[caleb_is_a_dharmabum]

+1 car reference.

I predict a quick death for XP after release of V.

[Opportunist]

There are rather few "good" reasons for the everyday user to buy Vista (unless it comes bundled with a new PC anyway). There have been many incentives to switch from NT to 2k (USB support), or even from 2k to XP (better support for a lot of hardware). But so far, the big "visible" incentive (aside of the 3D interface) is the DX10 support. Now, that's not something you can sell to a company. What for does a company need a component that mainly carters to gamers? Actually, most would love to NOT have it.

Also, there's the big black cloud of DRM that hovers over Vista, where pretty much nobody really knows yet just how dark it will be. Many people will abstain until that fog cleared, definitly something neither MS nor the content industry would enjoy. So, another incentive will be that certain content will only be available to you if you use Vista and its stronger DRM.

Another thing that doesn't bother companies too much. Actually, yet another incentive NOT to migrate, so your employees can't waste their time watching youtube.

What does bother companies, though, is support. So the faster support for XP ceases to exist, the faster companies will migrate. So, let the spinning start.

Whoopsie, already started.

Hardware and Software lifecycle

[suggsjc]

I'm not trying to flame here, but whenever a topic like this comes up there will always be someone posting about how they've had the same *nix/BSD box running for X years.

I do understand the concept of legacy hardware and software, and that if it ain't broke... However, almost EVERYTHING has a given lifecycle. I don't think that software should be any different. People are going to complain that M$ stops supporting their older OS'es (especially close to a new OS release) but honestly, how long should they be responsible for maintaining the code?
I hear the statement that "we paid for the software...so they should support it." In the open source realm, most people don't pay for the software, just for support and updates. So, in that same respect the people that bought windows paid up front for their support and maintenance, but how long should that be for? Is that something that should be included in the license...we guarantee to support this product for X years?

Sorry for the slight rant, but I know how people like to get all uppity about this stuff. But at least in this case I think it is completely justifiable.

What reasoning is that?

[guruevi]

"We don't allow Win 9x to connect to AD". It's not like there is a huge security risk for having AD run authentication for Win 9x. I can agree that you don't run AD on those boxes, but I have Win NT and Mac OS boxes connecting to AD. I can't change anything in the AD, I can just read stuff everybody else can read. Or is AD broken? In my company there are still Win NT 3 boxes standing around, they are firewalled...

Old versions of windows never die

[MECC]

They just slowly get virtualized....

Virtualization anyone?

[saleenS281]

This is exactly why we have VMware. Need to run an app for 98? Put it in a virtual session. Get all your *real work* done on the external OS, whether that be Windows/Linux/whatever. You turn on your network connection to the virtual machine only when you need to transfer files on and off of it. IIRC, you can also setup a firewall to block what can and can't get to that virtual machine... need ftp out? Only allow ftp. Most of this can be setup so even the most illiterate user can figure it out.

Re:Virtualization anyone?

[molarmass192]

... except if you need to make use of a specialized ISA/PCI card, that's where the weakness of virtualized hardware comes into play. FWIW, I have a vmware image of Win98, just in case, I've only ever used it to spy on USB traffic from Win-only USB drivers, and even that's been a while.

Fire drills

[AJWM]

Others say they also stay current to avoid headaches and fire drills.

Strange. I always though staying current was a headache and a fire drill.

(Heck, I still use 9.x on my kids' computers. Works fine for their software, and they're usually not on the internet. When they are it's behind a NATed firewall and using firefox.)

Point of Sale Systems

[Foofoobar]

On the day of the 11th, the day support for all Win 98 systems, I stopped by a Fedex and realized their POS systems (pun intended), were are win98. I let the guy know that Microsoft stops support for them and he said 'good luck getting corporate to upgrade'. At that point I realized that this was a POS system that was sold to them by another compny and that it is most likely that TONS of POS systems still ran 98.

I suspect that alot of companies at this point may actually decide to replace these systems with Linux based POS to save money and as a result of that, they will see the benefit of using Linux elsewhere as well. The big issue will be that these companies will have to upgrade all their terminals and hardware as well as all their software and potentially, if they just switched to Linux and a Open Source POS system, they could save MILLIONS.

Feel free to insert opinions here. I'm interested how others think corporate America will respond.

Re:stupid comentary

[Shipwack]

Well, we (a US Navy shop) have one machine running Win 3.1, and another that was "upgraded" to Win98... Though I doubt that as stand alone machines they qualify as being in a "high availibility enviroment". The machies run some specialised RF testing programs, and it just isn't cost effective to re-write the programs and/or QA the programs to run in a more modern version of Windows.

Is is obsolete beacuse it is old?

[plusser]

Just because a product is old, it does not mean that the product is obsolete. That is something that the IT industry needs to learn.

The Automotive industry is a good example. Suppose you bought a brand new car today, you would expect that you would be able to operate that vehicle for a number of years, after all it is a big investment. However, if the vendor said after 4 years that the engine could no longer be maintained and that it must be immediately replaced at your cost, you would not be very impressed. You would be tempted to perform your own DIY and install your own engine from a different vendor.

Thing is, Microsoft in recent years has tried to market a versions of Windows for embedded applications. When users of these operating systems realise that after 4 years that microsoft will expect you to upgrade a major piece of equipment as they no longer support the software it is based on, the customers are not going to be happy.

An old computer may run old software, but there is every chance that in every other respect that it may still be just as useful as a new one. The computer may have features that are no longer supported such as ISA cards or serial ports that are required to operate certain useful external equipment and embedded applications. In essence the cost of upgrading the computer operating system may be much greater than requesting that existing software is maintained. Unfortuately this is one area where Microsoft are running the risk of loosing the plot.

As for Microsoft saying that Windows ME is 6 years old and is therefore unsupportable, until 4 and a half years ago it was the latest operating system for home computers. XP isn't even 5 years old yet, but one thing is certain, if Microsoft imsists that I upgrade to Vista within the next 2 years, I will upgrade to Linux or OSX.

Re:Brings up the question

[Anne Thwacks]

We have several laptop machines running Win98 at work. Why? because they are used for engineering. They have floppy disk drives and serial ports, which are needed for engineering, and new laptops dont have them.

The old machines hare processors and memory which are far to small to run Win2k, and XP is too modern to be considered well enough tested for mission critical work :-)

When I have convinced people that Win98 is a security risk because its EOL'd so all the hackers know its a good virus target, these machines will have NetBSD installed. We cannot scrap them because we need them to support instruments that cost humungous amounts of money, and to run chronically obsolete tool chains to support products with a 30 year life span. - Yes its true - not everything with an embedded process or has a lifespan of 8 months, or even 8 years.

Think about it - some complex systems take two years to specify, and two to build, one for certification, then they take an age to get delivered and installed, possibly requiring a custom designed room, and then users take two years to learn how to use them, after that, people expect a 7 year _minimum_ product life. If you dont believe me, check out diagnostic equipment your local hospital, airport, rail depot, garage, etc.

Re:Article Summary...

[phillymjs]

Users may have custom software that does not work on new versions of Windows... could present IT challenges as Microsoft retires old products...

That's why Microsoft has such a hard-on for virtualization-- they want businesses to buy shiny new Windows 2003 servers and run, for example, their business-critical NT 4.0 legacy app that hasn't been updated, in a virtual machine on that server.

That's exactly why they bought Virtual PC from Connectix.

~Philly