You can add another factor here which is the observation that Linux networks are more stable, easier to administer, and less support burden than Microsoft networks. In the long run, support cost for Linux machines (especially large number of Linux machines) tend to be lower.
Here's a report from the European Commision that gives a few executive summary facts about Linux TCO:
- Licence fees account for 20% of IT budgets. Half of all respondents find this too high. This perception leads to increased future FLOSS use, especially among current non-users.
- FLOSS users administer 35% more PCs per IT administrator than non-users – FLOSS use appears to reduce administrator workload per PC, and IT departments with high workloads are more likely to want a future increase in FLOSS use.
Getting back to the original "Linux Wars" article:
Integration is another costly consideration. Customers attracted to Linux and other open source applications often say they're pursuing a best-of-breed strategy, according to Hollinger. "The harsh reality is that somebody has to be in the role of integrating those products," he says. "One of the things you buy in a Microsoft solution is product integration all the way up and down the suite."
Microsoft have done a good job selling the idea that they deliver product integration. In a limited sense, this is actually true but in a practical sense it is complete nonsense. The fact is that any given Microsoft suite provides lovely integration between programs in the same suite. Sadly, they do not tend to integrate with anything else -- not even with older generations of Microsoft products. This puts customers in the position of needing switch over to 100% Microsoft and to upgrade any older Microsoft boxes if they want reliable integration.
For instance, Beamer notes, "with a Windows-based solution, I know what my tool set looks like. With a Linux tool set, there are some dark areas out there."
Maybe for him there are black spots. Beamer seems to strangely believe that people implementing a Linux solution would hire Microsoft experts to help with that implementation.
Access to adequate support for Linux remains a question mark as well. "Linux has nowhere near the support structure Microsoft has in place, meaning that customers need in-house IT resources to maintain a Linux environment," Henson says.
This one is completely untrue, it can only be considered FUD (Fear Uncertainty and Doubt). From my long list of Linux-related organisations in the Sydney area it is clear to see that there is no shortage of Linux support. I'm sure that similar lists exist for any major city and there are probably even a handful of Linux gurus in typical country areas as well.
As for keeping in-house expertise... any medium sized business should keep some expertise on staff no matter which system they use. A business depending entirely on contractors for support is going to come unstuck. Contractors make a good resource but they are never an equivalent substitute for having in-house expertise. A staff member will be more concerned with long term continuity, will have interests that coincide more closely with the company they work for and will have better understanding of how the business works.
Having said that, it is still an excellent idea to hire contractors when you do find your staff getting out of their depth. It is also good to maintain ongoing relationships with contracting firms so you have the expertise available and don't get into a mad scramble. At any rate, other than for very small businesses, a support strategy not including any in-house IT resources is pretty damn stupid.
At the same time, Marshall warns, it's much harder to hire staffers who really know Linux.
It's pretty hard to hire staffers that really know Microsoft products too. You will find plenty who claim they do but what that means is they can reinstall when they get to something that they can't fix. In fact, it is very difficult to really understand Microsoft products because other than superficial user-level documentation, it is hard to obtain detailed information of the system internals.
In a general infrastructure support environment, the really critical requirement is someone who understands planning and testing. I've personally seen situations where a Microsoft expert has upgraded the local Microsoft email program with the result that all archived email messages were left incompatible and unreadable because they never considered a strategy for upgrading the archive data!
This sort of planning and contingency is basic stuff that a lot of so called "experts" are wont to overlook (until it is too late). Aside from such monumental stuff ups, most companies seem to judge their hiring strategy based on familiarity with system features rather than common sense and a good understanding of the fundamentals. If this is your preferred hiring strategy then you can find plenty of unix experts (from HP, SCO, Sun, AIX, etc) who have seen the writing on the wall, got a cheap Linux system at home and rapidly retrained themselves (because Linux is easy to pick up for anyone familiar with another Unix). Linux technical training is easier because the materials are easily available (including full internal documentation right down to the source code, plus heaps of books).
"Microsoft invests north of $6 billion a year on R&D. There is nobody in the Linux world" that does that, he says. "Whose economic best interest is it in to invest the extra $10 million or $50 million over some period of time to make sure that's a secure computing platform?"
Better ask the NSA why they decided to make "Security Enhanced" Linux. The sort of level of security offered by SE Linux is way beyond most users requirements but it is available right now from both RedHat and CentOS if you decide you need it. Better yet, you can run it in "warning mode" where it doesn't activate the full iron cage but at least it notifies you that things might be going wrong (and yes you can send those warnings to a separate log server to make it difficult for an intruder to erase their tracks).
What the R&D argument ignores is that Linux has already grown up faster than Microsoft and already performs better in most areas. The Open Source development process is vastly more efficient than the proprietary development process. The measure of money spent is not equivalent to the quality of the end product. You get less than what you pay for, sometimes just a little less, sometimes a great deal less. The Linux development process has not come for free, it took a bunch of people and a heap of effort but because those people exchanged ideas freely, tackled problems honestly, took user feedback onboard (most developers are also avid users of their own work) and chose technical directions primarily on merit rather than sales gimmicks or market buzzwords.
Henson says. "When the playing field is level, Windows outperforms Linux every time." Armed with the performance card, IT managers caught between rival Windows and Linux camps can defend choosing Windows by saying that "it's not a decision from my gut, it's not a religious decision -- the evidence showed that Windows was better," says Henson.
I think we all remember the Mindcraft tests. The final result was khttpd which is the fastest webserver around for static pages.
I've run Microsoft SQL Server 2000 against the Open Source MySQL of similar vintage (back around 2001). This was using fairly standard single CPU servers which were the most common hardware at the time. The test involved summing and grouping large volumes of non-indexed network data. The sort of thing that requires raw speed rather than clever query optimisation. MySQL handled it excellently, always at least twice as fast as Microsoft SQL Server 2000. That was the last thing I did with Microsoft SQL Server so they have probably made some changes since then.
I've heard lots of other test results. One that sticks in my mind was how long it took Microsoft to get Hotmail switched over from FreeBSD to Windows because they couldn't achieve the necessary performance. This article is Microsoft's discussion of the Hotmail conversion, which doesn't tell you that it took them years (and several aborted attempts) to get it switched over. Another big eye opener is that Google prefer to use Linux clusters. I'm sure a lot of Google's strength is clever algorithms but raw speed isn't something they are going to knock back either.
Addition 2006-01-14:
Looks like Paul Murphy also doesn't hold much faith in Microsoft's performance comparisons.Si Chen's discussion of the "Linux Wars" issue examines financing Open Source development.
Such a development model can be highly efficient but would not address some important market needs. For example, it cannot help non-technical consumers who cannot develop the features they want themselves. (Think about 100,000 users with $10 each to spend. That's $1,000,000 of development which would probably never find its way into open source.) It also fails when, based on the laws of probability, there are simply too few users to create a large pool of developers. Finally, it would also not work very well when the users themselves could not envision the features they would actually need.
The Bounty System is providing a method for users with a bit of cash and insufficient technical know-how to put money into development of the products and/or features that interest them. It provides a competitive marketplace where the first developer to produce a result that meets specification can collect the bounty. Most bounty postings are larger than $10 because with such small sums of money, the transaction overhead is large and the logistics of collecting all the money become too much of a burdon. On the other hand, how many Microsoft products are on sale for $10? None I would say.
As for users being unable to envision the features that they need... On what basis does Microsoft have better ability to tell the users what they need than the users themselves? Yes, I'm willing to accept that it is theoretically possible for some well trained expert to make an insightful revelation but when we get down to practical reality -- how often does it happen? Microsoft's main expertise is marketing. This is what brings in their money. If Microsoft make a slightly better or slightly worse product, how much does this impact their sales figures?
Most buyers of commercial software don't actually verify that its features are bug free or check out its support lines. Instead, their "due diligence" consists of making sure that there are other users using the software, including, most importantly, their golf buddies.
This is completely true. There is a chicken and egg problem where everyone recommends only what they know which comes from other people's recommendations which is only what they know. No one wants to go out on a limb and recommend something new and different. It is also a restatement of an older and more famous quote:
And it ought to be remembered that there is nothing more difficult to take in hand, more perilous to conduct, or more uncertain in its success, than to take the lead in the introduction of a new order of things. Because the innovator has for enemies all those who have done well under the old conditions, and lukewarm defenders in those who may do well under the new. This coolness arises partly from fear of the opponents, who have the laws on their side, and partly from the incredulity of men, who do not readily believe in new things until they have had a long experience of them.
Yet despite this inbuilt resistance to change,
one thing we can be absolutely sure of is that change will occur.
This work is licensed under a Creative Commons License.