Joe Barr

Subscribe to Joe Barr: eMailAlertsEmail Alerts
Get Joe Barr via: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Related Topics: PR on Ulitzer

PR: Article

Seeing through the Linux-Windows TCO comparisons

The conditions & assumptions behind prominent TCO claims are unlikely to be found in reality.

(LinuxWorld) — As happy as I am to see something finally displace the disgrace that TPC-C "benchmarks" have become over the years, the new favorite faux-analytical geegaw that Microsoft's marketing droids are giggling over isn't any better. I think you know what I'm talking about: it starts with a T, ends with an "Oh!" and spells trouble right here in River City. That's Austin, dude. Texas.

On the face of it, Total Cost of Ownership (TCO) sounds like it should be a valuable tool to project cost-estimates of various software solutions. There is certainly truth to the notion that purchase-price alone doesn't determine total cost. The problems with published TCO studies begin and end with the fact that their allure is only skin-deep.

For one thing, TCO is like fine wine: it doesn't travel well. What may be true in one situation is reversed in another. What gets trumpeted as a universal truth ("Windows is cheaper than Linux") may or may not be true in a specific case, but it is most certainly false when claimed universally.

Many variables come into play. If you are an all-Microsoft shop with no IT employees skilled in other platforms, the cost of your migration to Linux is going to be greater than a shop whose assets include cross-platform skills.

Highly skilled Windows professionals are going to have an easier time migrating than those who struggle to master its click-and-drool mentality and shy away from the command line like it's calculus.

These factors — and others that may be invisible to the outside world — drive the costs that comprise TCO in that specific environment: planning, training, implementation, maintenance and administration.

When you read about a TCO study in the press these days, you're not reading news. You're reading marketing material. More likely than not, you're reading a report sponsored by the vendor. If the comparison is against products from another vendor, the sponsor is the one whose ox was not gored. The recent IDC report proclaiming that Windows is cheaper (in some cases) than Linux is an excellent example.

(Editor's note: Both IDC and LinuxWorld are owned by IDG.)

Because Microsoft funded the study, they got to choose the parameters. They chose Windows 2000, and if they had used products sold under Licensing 6.0, the results would have been much different. Ditto the stipulation that Windows 2000 be run for five years without an OS upgrade. In real life, Microsoft is doing everything it can to get people off of Win2K and on to the more expensive spread. Even Microsoft admits that would mean higher TCO. Only in the twilight zone of TCO mania is a five-year upgrade cycle acceptable.

The bottom line is this: Those who pay for research decide what data goes into the calculation and, perhaps even more importantly, what data doesn't. You can see why TCO is becoming so popular with marketing types; it's rife with opportunities to mislead.

Enough of that. I am not here simply to bash published TCO reports. Like my longtime favorite IT journalist Nicholas Petreley said, "slamming the conclusion that Windows has a lower TCO than Linux is like debunking the claim that there's a statue of Elvis on Mars. Why bother?"

He's dead right. What I really want to do is to help those of you who want to undertake your own TCO study. To that end, let's take a look at three current or recent real-world scenarios in order to learn what elements should be included in your calculations.

First, let's go back to school. Quoting from Steve Duin's story in The Oregonian earlier this year:

Say it ain't, Joe... and Steve and John and Scott and the rest of the computer tech supervisors at the 24 largest school districts in Oregon and Washington.

At the busiest time of the year for those districts, Microsoft is demanding that they conduct an internal software audit to "certify licensing compliance." In a March letter, the software giant gave Portland Public Schools 60 days to inventory its 25,000 computers.

"Which," said Scott Robinson, the district's chief technology officer, "is a virtual impossibility."

Was that legal? Lawyers say yes. Ethical? Not in a million years. Typical? Unfortunately, yes. And very. Redmond has a blackbelt in abuse of its monopoly power, and it doesn't hesitate to wield it against customers as well as competition. Forget for a moment the fact that the Bush administration sold Microsoft a get-out-of-antitrust-jail-free card. Think about all the things Microsoft did to get in hot water with the DOJ in the first place.

Similar events have happened around the country. According to ConsultingTimes.com, Microsoft put Los Angeles schools between a rock and a hard place:

The Los Angeles city schools fell even deeper into the licensing-compliance pit in 1996, when an audit revealed several hundred unlicensed copies of Macromedia and Microsoft products. At $150,000 per violation, the potential cost was $19.8 million, plus cost of audit and legal expenses. The schools had already spent $8 million for technology during the 1995-96 school year, but they had obviously not budgeted enough to make sure their licenses were provable.

That last sentence is worth reading again: "The schools had already spent $8 million for technology during the 1995-96 school year, but they had obviously not budgeted enough to make sure their licenses were provable." Sure sounds like TCO material to me.

After the letter

After you get "the letter," whether it comes directly from Microsoft or its thugs at the BSA, you have two options:

  1. Bite the bullet and negotiate a settlement
  2. Sign up for the infamous Enterprise Agreement

You might be surprised to learn that the first option has shown to be more cost effective than the second. Let's look at two such cases: one chose Door 1 and the other chose Door 2.

As reported last month in our sister publication InfoWorld, the BSA and U. S. Marshals raided Ernie Ball, Inc. two years ago. Its IT department was shut down as its computers were searched for "illegal copies" of software. The raid itself was said to be the direct result of a single phone call from an ex-employee to the BSA.

Ernie Ball, Inc. negotiated with the BSA. The haggling began at $300,000 but was whittled down to $90,000. But it is what Ernie Ball did afterwards makes it interesting from a TCO point of view. Today, just two years after the raid, the company is Microsoft-free. Almost everything is now open-source. Not just the servers, but the desktops too. As a result of the migration, Ernie Ball, Inc. estimates it saves $70,000 to $100,000 a year, which is not bad for a company with 70 users.

Here in River City

The City of Austin had a similar situation, although it didn't begin with armed marshals shutting down its computer operations. Austin performed a "voluntary" audit at the request of Microsoft legal rather than the BSA. Although the City says it wasn't completely due to the audit results, it admits that the findings did influence its decision to enter into an Enterprise Agreement version 5.0 at about this time last year.

Earlier this year, I requested, under the Texas Open Records Act, a copy of the agreement and copies of invoices for new computers purchased by the City of Austin since the Enterprise Agreement went into effect. The City balked at providing the invoices. Brad Norton of the City Attorney's office told me they would not waive the fee for preparing copies of the invoices because it wasn't in the public interest. Perhaps the City is embarrassed about paying for Windows twice. If they're not, they should be, especially since the City is in the middle of a severe budget crunch due to a $40 million shortfall in revenue.

Because it is most certainly in the public's interest that they learn how their tax money is being spent, I paid the fees for copying the invoices out of my own pocket. The invoices I received from the City of Austin were all from Dell Computer. Each invoice itemized the hardware and software included with the system, but only a total price was shown. Based on the invoice, it is impossible to tell how much each hardware or software item cost.

October 3, 2000, the state of Texas' Department of Information Resources signed a "Microsoft Government Enterprise Agreement 5.0," which is a three-year agreement covering state and local governments. (See links in the resources section below.) The city of Austin elected to participate in the agreement, and finalized the details in late 2001.

In spite of Austin's participation in the pricey Enterprise Agreement, the invoices showed clearly that the City is continuing to pay for a copy of Windows on each and every new desktop system purchased. I also learned that Dell Computer is dead-serious about not revealing how much it is charging the city for Windows. Under the terms of the Enterprise Agreement, the City pays for Windows again when it "trues-up" at the end of each year.

I called Dell public relations and asked if they could give me a ballpark estimate on the cost of Windows 2000 and Windows XP Pro. After asking why I wanted to know, they promised to get back with me. They didn't, but Microsoft did.

A gentleman from Microsoft declined to discuss the licensing fees it charges computer manufacturers. He did seek to reassure me, however, explaining that Microsoft expects to be paid twice by the City. Well, wait... he didn't use those exact words, but that's what he meant. He told me that under an Enterprise Agreement, Microsoft assumes that when a computer rolls in an Enterprise Agreement customer's door, its hard disk will contain a legal copy of Windows.

I asked him to point out to me where in Enterprise Agreement 5.0 that assumption is stipulated. Alas, each agreement is different, he said. Besides, Microsoft treats each as confidential and simply won't discuss agreements with anyone but the customer. Fair enough, but it didn't matter, as I had a copy of the Texas agreement right in front of me.

I combed through it and could not find such a clause. However, I found something even more interesting. In the preamble of the agreement where Microsoft defines terms and conditions, I found Microsoft's definition of desktop computers encompased in the agreement. It defines "qualified desktops" as:

personal desktop computers, portable computers, workstations and similar devices, which are used by and for the benefit of an enrolled Affiliate or any Affiliate included in its enterprise and which meet the minimum requirements for running any of the enterprise products.

Single-use, special-purpose computers and servers are excluded, but clearly a desktop running Red Hat 8.0 or Mandrake 9.0 is a "qualified desktop." This is important at the end of the year, because that is when the City has to "true-up."

According to the agreement, page 4, section 3.c: "Each enrolled Affiliate must determine the current number of qualified desktops in its enterprise at each anniversary of the effective date of its enrollment and at the expiration or early termination of the enrollment. If the number has increased, the enrolled Affiliate must submit a purchase order to its reseller to license those additional qualified desktops..."

In other words, Microsoft doesn't care if the qualified desktop is running OS/2, BeOS, Linux, or FreeBSD. It gets paid for each desktop's theoretical ability to run Microsoft software. A law school Contracts professor could use this clause as an example of drafting that's either sloppy or overreaching.

I asked Rick Wallingford, owner of one of Austin's oldest local computer shops, if he could sell the City similar desktops without Windows installed. Certainly, he said. Alternatively, thanks to Austin's participation in the Enterprise Agreement, Wallingford said he could master and install all the products it licensed under the Enterprise Agreement, including the operating system, without charging for the software, assuming Austin included those PCs in its inventory of "qualified desktops."

If Wallingford is correct, and my understanding of the documents suggest he is, Austin buys Windows twice for each new PC it purchases. I asked Brad Norton — a lawyer from the City Attorney's office — if the City believed the Enterprise Agreement required them to pay twice. He had no comment.

Austin and TCO

Strange that such service is available in Austin but is ignored by the City. It's almost as if the City of Austin IT department wants to single-handedly disprove TCO studies showing Windows is cheaper than Linux by wasting as much money as possible.

Just for the sake of having a number, let's assume the city is paying an average of $125 for Windows on each desktop PC it buys. Extrapolating the number of PCs purchased in the four-month period between January 25th and May 22 of this year, the City of Austin will pay $91,000 this year for Windows licensing. There is no reason to think that number will do anything but grow the next two years of the contract.

Final tabulations

What can we learn about TCO calculations based on these events? I don't know about you, but five items that are probably not part of Microsoft-funded TCO studies leap out at me. Here they are:

  1. You need to add the costs for legal advice prior to signing any agreement with Microsoft. Otherwise you may be hit with demands that are impossible to comply with, like those made on the school districts in the Pacific Northwest.
  2. You need to add the costs for the purchase and maintenance of a software-licensing inventory system, as well as all the costs associated with keeping it up-to-date throughout your enterprise. This is true even with an Enterprise Agreement, because you will more than likely add applications to some desktops that are not covered under the agreement.
  3. You need to add the cost of insurance to defray the expenses associated with compliance with BSA or Microsoft requests for a "voluntary" software inventory.
  4. You need to add the cost of insurance to cover the $151,000 per-incident liability incurred if a proof of purchase is misplaced.
  5. Unless you're careful, you may need to add the cost of Microsoft "double dipping."

Is TCO a valid yardstick for evaluating your IT solutions? The answer is a firm maybe. But it is only as good as the data and the template used to calculate it, and every company's situation is going to be unique. Do not — REPEAT — do not trust a vendor-funded study as the basis for plotting your IT strategy.

More Stories By Joe Barr

Joe Barr is a freelance journalist covering Linux, open source and network security. His 'Version Control' column has been a regular feature of Linux.SYS-CON.com since its inception. As far as we know, he is the only living journalist whose works have appeared both in phrack, the legendary underground zine, and IBM Personal Systems Magazine.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.