This is useful in various contexts. We keep hearing from various quarters about how well they are doing, “beating” the competition. Well, at the end of the day, its what people do, not what they say that matters.
Top-Level Server Market Findings
* Linux servers posted the fifth consecutive quarter of accelerating revenue growth[ed emphasis mine], with year-over-year revenue growth of 19.0%, for a total of $1.8 billion in the quarter. Linux servers now represent 13.6% of all server revenue, up from 12.1% a year ago.
* Microsoft Windows server revenue was $5.0 billion in 2Q07, showing 18.7% year-over-year growth and gaining 4.0 points of revenue market share over 2Q06 and comprising 38.2% of all server revenue in the quarter. Windows servers account for the single largest segment of spending, by operating system, in the worldwide server market.
Ok, lets work a little math:
1.8B$/5.0B$ US is … 0.36
That is the newly purchased Linux server market is 36% the size of the newly purchased windows server market, and growing.
Notice the wording I used. I chose the words carefully.
There are quite a few (no, I don’t have numbers) cases where an admin will wipe windows off a server and install Linux on it for a point solution. To address a need.
Why am I noting this? Three reasons
- This is not accounted for in the data above. There is no measure of continuity of shipped OS remaining on the system over its lifetime.
- As far as we know, the reverse doesn’t happen with any significant frequency
- The data also does not take into account new units shipped with windows licenses, as it is next to impossible to get some vendors to ship you the server you want without the specific OS they want to ship with it. An estimate of this may be found online at eBay, where you can buy valid/original windows licenses/CDs/COA’s by the boatload, usually from customers who have been shipped licenses/CDs/COAs, and do not need/want them, and need to monetize the extra cost they paid to the server vendor for shipping them the unwanted licenses.
Without firm data, it will be hard to estimate the impact of these. Suffice it to say that the Linux numbers IMO represent a lower bound on that market. OTOH, the same arguments suggest that the windows represent an upper bound on the size of that market.
Remember that there are some people who like to argue that there are few enough people with Linux experience to administer them. The numbers rather badly belie their assertions. For ever 3 dollars spent on a windows server, it appears that about 1 dollar is spent on a Linux server. As I am sure Google isn’t the only purchaser of Linux servers out there, there may be more than a few experienced Linux admins out there. If you ignore that you can configure a Linux server with less “stuff” in it (and therefore at a lower price) than an “equivalent” windows server, and simply take the dollar volume ratio as being equivalent to the numerical installed base ratio, you get a 1 in 3 ratio again. But as indicated, this is likely to be the lower bound for a number of reasons.
The data suggests non x86/x86_64 is on the decline. It suggests that Linux and windows are growing rapidly. It suggests that someone out there is buying those pesky Linux boxes.
Which also means that someone is administering them.