Latest Ballmergram Bashes Linux TCO

Phoe6 writes "Microsoft chief executive Steve Ballmer has used the software giant's latest executive email to stoke up Microsoft's fight against the rise of Linux. The 2,600 word missive was titled 'Customer focus: comparing Windows with Linux and UNIX'. In it, Ballmer repeated the key themes of Microsoft's controversial Get The Facts campaign. Zdnet has its report here." Linuxworld also has a story.

Comments

Ballmer wants to hear from you

[scupper]

Excerpt from last paragraph of Ballmer memo:

If the evidence at our www.microsoft.com/getthefacts Web site doesn't sufficiently convey the benefits and value of the Microsoft platform, we want to hear from you so we can work even harder to get that information to you.

I can't wait to read the the response to his invitation.

Groklaw

[julesh]

Also worth reading the groklaw article on this, which is available here.

So does the FDIC

[TykeClone]

http://www.fdic.gov/news/news/financial/2004/FIL11 404a.html

MS is sweating

[nels_tomlinson]

Monkeyboy is sweating. Unfortunately, that's nothing new.

Seriously, this is just the marketroids doing their thing. When the accountants start warning about threats from Linux, we know there's a real threat. Linux is getting mention in the latest annual filing, too.

Windoze in .edu? Bullshit on lower TCO!

[deunan_k]

With an educational discount, Windows 2003 server costs my department $142 (sure, it's more in a commercial environment), I can install it, set it up, run updates on it once in a while, run the baseline security analyzer on it, and forget about it. Give me a linux that does all this in an easy to use manner, and I'll switch. Sure, I could use apt and the others, but it just takes too much time, and you have to worry about various dependencies and what not.

At $142, that's $142 more you have to spend compared to FOSS solutions. What you've described, proved either that your educational institution is filthy rich and caters only for the rich and snobs, or you're just plain lazy.

Most educational institutions, whether state-run or even privately operated (esp. private with visions of education rather than for profit), are almost always tight budget! This is especially true in third world countries! That is why various bodies such as SchoolForge (and their Case Studies), K12OS, Moodle, OpenSourceSchools, KDE Edutainment Project and a lot more others are being founded and.. surprise! Thrives!

Personally, I love the K12LTSP Project. A branch out of the K12OS Project, which when deployed properly throughout the campus, can provide access to all students to high-grade apps in a very stable environment. Access from any terminal in any labs, authenticating via NIS, LDAP or whatever you prefer and access your mail accounts, website or whatever. With backend support tools available such as MySQL or PgSQL and PHP/Perl (okay, maybe that's a bit far out, but I've met 12 year olds who can code!)

Software cost? $0

Re:maybe the TCO is lower

[julesh]

Users are limited to 16 groups??

32, I believe. Still, it isn't perfect, and we should perhaps look at ways to improve it.

Only one group can have permissions applied to a file?

Not true. All major Linux filesystems support POSIX ACLs now, enabling you to apply whatever permissions you like.

And no group nesting allowed?

What are the security benefits of allowing this? Personally, I am not aware of any, as I believe whether it is allowed or not the systems are actually equivalent -- it is merely an implementation detail that should be ironed out by any reasonably well written management system.

You can mod this comment down, but you can't propose a security system like THAT to a company interested in protecting their assets. WAKE UP SLASHDOT.

Even without ACLs, it is more than adequate for 99% of companies. Hell, most of them wouldn't want to spend the admin time required to manage anything more complex.

TCO

[TheFlu]

I appreciate the fact that it's going to cost a company some money to switch from Windows to Linux, mainly to hire competent administrators (and if you're thinking of doing so, hiring competent administrators is an absolute must).

However, I also appreciate the fact that said company is never going to have to pay for a software and/or operating system upgrade ever again. This is called smart spending. You shell out money in the short term to save significant amounts of money in the long term.

Speaking from experience here, my company has switched every machine in our office to Linux, both servers and clients, and we've saved a bundle in the long run by doing so.

Re:Windows TCO

[boodaman]

Why does there have to be a 1-to-1 relationship between what you can do with Linux and what you can do with Windows?

Have you considered that "DFS" might be a solution looking for a problem?

For example, from the DFS FAQ (http://www.microsoft.com/windowsserver2003/techin fo/overview/dfsfaq.mspx):

For example, if you have marketing files on multiple servers in a domain, you can use DFS to make it appear as though all of the marketing files are on a single server.

So why would you have marketing files on multiple servers in a domain? This is a symptom of a larger organizational problem: the balkanization (if you will) of storage by most companies. The way most companies solve their storage problems is by buying another server, typically with a couple hundred GB of storage.

Then they have issues like "Susie, if you need the marketing files from 2002, they're on the server called HARPO, but if you want the ones from 2003, they're on GROUCHO because HARPO ran out of space. And the 2004 files are on MOE. Or maybe CURLY, I can't remember, ask the help desk."

The issue isn't "Linux doesn't have DFS", the issue is "most companies manage storage (and knowledge) poorly".

Many corporate storage problems are a result of poor workflow, poor process management, and an insatiable need by most management grunts to cover their asses by saving EVERYTHING no matter what instead of assessing what really needs to be saved, and organizing it in a way that makes sense to the organization.

The solution to the problem isn't "use Windows because it has DFS and UNIX doesn't" but "disconnect storage from processing". Don't buy more disk by buying another server...just buy more disk, or better yet, figure out why you're using so much disk and solve THAT problem instead.

It is easy to get caught in the "server == disk" trap, because you have companies like Dell selling "servers" for $1000. That's great for the short-term, but the long-term costs go up, up, and up as complexity increases and the need to have more admins to manage that complexity increases along with it.

The long term solution is to understand that you will ALWAYS want more disk, and plan accordingly by buying a "real" server that can accept external arrays, etc. or better yet, buying a filer solution from Hitachi, Network Appliance, or similar. The short term cost will be higher, but its just hardware so you can depreciate it anyway, and the long term costs will be tremendously lower. And your DR (disaster recovery) processes will be much cleaner and more robust.

In the spirit of Ballmer's e-mail, the TCO on a filer solution is much lower than a corresponding TCO for managing several, or dozens, or hundreds of servers, each with a couple hundred gigs of storage on it (not to mention server OS licensing costs, archiving software costs, etc).

Heck, if an organization is REALLY smart, they won't even use Explorer-style file management...they'll have an intranet where people search for what they want in a browser, with the results coming from a DB and all they do is click on the "download" link which retrieves the file for them. They never, ever have to know where the physical file even resides.

Re:From Linux to Windows

[f()rK()_Bomb]

I would second this as i worked in a pure 2000 domain but if you actually think about uptime exactly how often do win2k servers reboot? We were generally rebooting the things every couple of weeks due to patches. Leave one up for more than 60 days and things start to get wierd. For instance our DCHP server decided to stop giving out addresses. Reboot and it was fine. I would definately consider this as unstable.