Our Response to Microsoft’s Misleading IDC Numbers

amcpherson's picture

Peter Galli printed a fair article questioning Microsoft’s slicing and dicing of raw IDC numbers.

Here is the full text of my response to his original article:

Fundamentally this particular study will over-count Windows share and undercount Linux. Al Gillen at IDC, who we have a lot of respect for, says this himself in your article. Why is Linux so under-counted in this research?

1. This study looks at pre-installed servers. Many people buy bare
servers and install Linux on them. Many people are recycling old
Unix or Windows servers and installing Linux on them. These
servers are not counted in this study — it’s no fault of IDC,
it’s merely impossible to do so.

2. This study only looks at x86 servers. Huge Linux growth on other
architectures like mainframes, P-series, etc has been widely
reported by IDC and others. Of course when you discount other
architectures Linux does well on and focus on the /one/ Microsoft
excels at, you’ll get highly favorable rates for Microsoft.

3. This study fails to consider the impact virtualization has had on
server shipments. Customers are optimizing boxes they have already
purchased via virtualization. Since Linux is considered the better
platform for virtualization, this impacts its numbers even more. I
don’t think customers are complaining that they are getting more
out of the investments they have already made.

4. This study was confined to the US market. Linux growth worldwide
is even more impressive than the numbers cited here and would
certainly contradict the overall point made by this argument.
Again the data is diced in such a way to paint the most favorable
scenario for Microsoft.

Every piece of research, no matter the talents or commitment of the researcher, has a margin for error. Because the margin for error is higher in this study for the reasons above, it’s a stretch to take a few quarters and proclaim it a trend. There really is no empirical data that we can find that justifies the “low hanging fruit” trend cited in your article.

Growth rates are always sensational so I’m not surprised you’re quoting them here, but let’s put it into perspective. The Linux server market is now a $7.7B business - it’s hard to grow big markets like that at 30% indefinitely especially when the overall server market has been slowed in recent years. Of course Linux growth will slow as it becomes more and more predominantly used.

As for Mr. Hilf’s experience with Linux only be used in web server and high performance computing workloads, I don’t agree with him (even though I am happy to see him talk about Linux used in HPC environments – this is quite a change!). We also talk to many Linux customers around the world and find that Linux is being used for ERP/CRM, application development, file and print and many other workloads. Just ask Novell and Red Hat for customer references. I would be interested in hearing from IDC what they are seeing from their data on this issue or work with them directly to find this out. I think this would be superior than relying on Mr. Hilf’s analysis.

But make no mistake: we’re not Pollyannas — Microsoft is doing very well. They have a huge incumbent position. They have unbelievable resources they put toward marketing (this being just one example.) They have kept interoperability information propriety (such as in file and print) to hinder competition. They are a hugely successful software company who will do anything it can to squash the competition, in this case Linux. Linux’ success against such a well-placed, well-stocked competitor has been phenomenal so far, but we all certainly need to keep up the innovation if we want to continue to put pressure on them. Microsoft will only increase its offensive against Linux, especially as Linux and Open Office threaten its two cash cows. We are prepared for that and are confident that over time customers want the freedom of choice that only open source can provide.