Saturday, September 16, 2006

The problems with reliance on Microsoft

Here we go now. One of my pet peeves over the past couple of years has been the reliance on Microsoft and Vista. If you are reading this, you've probably done see the headlines about how the computer industry is in turmoil because Vista will ship in 2007. You've probably seen the by-lines about how the industry is dependant on the Vista software eco system. You've probably seen the editorials about how Microsoft's late shipments has hurt computer sales.

What you may not realize is that these reports are utter bunk. Here's why. If you are not involved in the stock market, find someone who is. What I want you to ask them is if you had a set amount of money, should you invest it only one company, or several companies. Any good stock broker will immediatly advise you to... what? Diversify. Put your monetary holdings in multiple companies. Put some money in "guaranteed" stocks like Power Companies, and put some money in high-growth companies, like web-start ups.

It's common sense really. Yes, it's risky to bet on only one company, and if they do break through, then you do stand to regain quiet a lot. Example, those who bet on Microsoft back in the 1970's and 1980's. Those who bet on Google moving into 2000 and onwards.

But, if the companies fall, say like Pets.com or Enron, where is your money? What did you get back?


This is the lesson that vendors who worked with Pure Unix learned a long time ago. Now, for those who haven't worked with pure Unix before, you may not be aware of what Vendor Lock-In really is. Vendor Lock-In is when you can not upgrade, cross-grade, downgrade, or do anything with the hardware and software you use without paying a specific vendor of that product to do the work for you, or provide the software for you.

What happened with Pure Unix systems is that many vendors would change specific code relating to the kernel or other items, then link their program into those changes. The result was that you couldn't change the Operating System, the Unix kernel itself, and still be able to run the program. By the same token you couldn't change the program unless your new program was built for the customized kernel already in use in the old system.

So, if you wanted to upgrade your software to the latest version, you would have to not only buy the new software you use, you'd have to buy a new Operating System for it.

Now, I'm not picking on IBM with this, they will just set here for the example. IBM's Pure Unix is sold as AIX, or Advanced IBM Unix.

Lets say you bought a mainframe from IBM, and this mainframe had AIX version... 2. And it had, IBM Finance Manger Suite version... 3. The total mainframe cost is going to hit you for $10,000 at the start.

You use this software for about a year and you find that IBM Finance Manager Suite version 3 just doesn't do it all for you. During this time IBM has released their new version of the mainframe and software package you are using, for $10,000. You really, really, really need to upgrade to IBM Finance Manager Suite version 4. IBM is willing to sell you the new version for only $1,000. Great, that's 10% of the cost of a completely new mainframe.

You get IBM Finance Manager Suite Version 4 in and try to install it. Only, it won't install. It's not compatible with AIX Version 2. You need AIX Version 3, which is a $5,000 upgrade fee. Alright, so you consult with your board members and decide you are going to shell out for the Operating System. You pay $5,000 more for AIX Version 3. You get it, install it, and it now runs.

Only... it runs slowely. Very slowely. Your mainframe isn't fast enough to run it. Great, now you need a new Mainframe. You talk with IBM sales, and they agree to cut you a deal. You've been a good customer, so they'll swap your old mainframe out for the newest and fastest version, for $8,000. Again, you go back to the board, and they say... yeah. Lets do it. So you drop out $8000 and now you have the new mainframe, and it works. Now everything is right.

Now read that again and add how much you spent. $1,000 + $5,000 + $8,000 = $14,000

You not only paid for another server, you gave a healthy $4,000 bonus to IBM.

Now, today, collectively, you think, we wouldn't put up with that. If a sales person tried that, we'd go elsewhere. What if you couldn't? What if SGI's IRIX, or HP's HPUX couldn't run IBM Finance Manager Suite... at all? What if you went and got HP's Mainframe, but it couldn't do half the stuff IBM's mainframe did? Where would you be then?


Welcome to the concept of vendor lock in. Most Unix Guru's are well familiar with these examples, having either lived through them, or performed them.

This is one of the reasons why Linux took off. It was completely documented. You didn't, and don't, have undocumented code or secret run times that only the manufacturer knows about.

Think about it for a second, then skip over to http://www.distrowatch.com Take a look at all the Linux Distro's listed there. Are you aware that just about any program compiled for Linux can run on all of those distrobutions? Without the need for modification?

When you get back here, think about this: Less than 10 years ago it was common that if you had AIX, you couldn't use programs built for IRIX or HPUX, and vice versa. Sure, programs that strictly adheared to the POSIX standards might run, but more than likely, they would need to have some sort of porting done to account for the differences.

Alright, so how does this apply to the launch of Vista?

It goes back to Vendor Lockin, and it goes back to diversity. Another real life example on the way to the main point was looked at by El'Reg. You can find the link here: http://www.theregister.co.uk/2006/09/14/amd_market_cap/

Vendors who invested in AMD hardware in addition to Intel hardware saw their share price rising. Those who only invested in Intel hardware saw their share price drop.

It wasn't exactly a no brainer idea to invest in both AMD and Intel. It made sense. Either company could fail, or could suceed. HP is a shining example of this. Sun, not so much since they have their own architecture in SPARC.

But the point is, the smarter, and more profitable hardware manufactuers realized years ago that they could not rely on one vendor alone to fulfill all their needs and all their production lines.

Dell, the poster child for successful vendor dependance finally went dual vendor earlier this year, to the cheers of almost all with the technology crowds. Why? Because consumers now had a choice from Dell. They now had parity for the desktop market. Even Dell admitted it was a mistake to rely on one vendor so much. You can find that admittance here : http://searchdatacenter.techtarget.com/originalContent/0,289142,sid80_gci1215577,00.html


Alright, so, what does this have to do with Vista?

Isn't it obvious?

Why are software vendors relying on one company to deliver everything?
Why are hardware vendors relying on one company to deliver everything?
Why are consumers relying on one company to deliver... everything?


Here's the problem with the reliance on Vista. All the empiracle evidence out there, in software, stocks, hardware, whatever, tells us it's a bad idea to bet everything on one product or one company.

Thing about AeroGlass for a second. Sure, a 3D display is nice... if it weren't for the fact that Novell is already there... with XGL : http://www.novell.com/linux/xglrelease/

Sure, DX10 is a nice advance in 3D features... except OpenGL was there... with Version 2.0

Care to guess when that was?

Version 2.0 - October 22, 2004
http://www.opengl.org/documentation/specs/version2.0/glspec20.pdf

OpenGL is already to 2.1, released June 30th, 2006
http://www.opengl.org/documentation/specs/version2.1/glspec21.pdf

DirectX 10 isn't even out for consumers yet.

My problem is that OEM's, like Dell, like Gateway, like HP, like Emachines, and others, have been hooked on Microsoft. They've gotten tunnel vision on what Microsoft can provide.

The problem is, Microsoft cannot provide everything. They can't do it all, even if they don't want to admit that. Just look at the software ecosystem. How many people buy an additional Anti-Virus? How many people use a different browser? How many people use a different media player? How many people don't use Microsoft products?

When I look at the computing industry, I really wonder where the OEM's that put computers in retail chains have been these past year. They know what customers want in a desktop.

Hey, I worked Phone Support for Cox Communications for 3 years. I got indulged with calls by people who told me how it "should be" for them, and how the product they used would be good for them? Where did I refer those customers to? www.opera.com and www.mozilla.org

I know for a personal fact that these callers are telling the support personal they call how things should be. And while a lot of the suggestions are junk (For example, having your passwords listed in a pop-up window), there are some suggestions that could be put to use (e.g. Tab functions, now standard in all browsers)

Thing is, OEM's have been tunnel visioning past entities like Mepis and Linspire that are seeking to change the very way consumers approach and use their computer. They've ignored ones like Symphony that are trying to change the very concept of a desktop.

Why? OEM's are the ones who most valuable to these projects. It's in the interest of Dell, HP, Gateway, and whoever else to invest in Linux systems just as much as they invest in Windows systems.

Think about it. What if Dell had invested in Linux for the consumer desktop? Okay, big deal that Microsoft's Aeroglass won't be here till next year. Dell could already be shipping XGL enabled desktops, TODAY.

Not next year.

What if ATi and Nvidia had invested in Linux drivers in the same way they had invested in Windows Drivers. OpenGL 2.1 support could already be implemented. We could already be playing Quake4, Half-Life 2 Episode 2, and F.E.A.R. Extraction Point under OpenGL 2.1 and get all the benifits, without having to wait till next year... when DX10 arrives.

That's my problem with OEM's and Vista. Computer manufactures are missing out on future promises that here today.

They are locking their customers in, not giving them a choice on what they can buy. The Dell's, HP's, Gateway's, and otherwise, today, are just as guilty of Vendor-Lock-In as the Pure Unix providers of yesteryear.

And I don't think that's right.

No comments: