I recently published a paper on one of my other sites that examined the role virtualization is playing in the market for Linux software and server. I've gotten enough positive comments that I'd like to summarize it here.
Analysts and their views
Analysts base their market sizing and market share information based upon different metrics. What they see, of course, depends upon what they're looking at! Here are some of the different metrics that are relevant to an understanding of an operating system market.
- Overall operating environment shipments
- Overall operating environment revenues
- Server operating environment shipments
- Server operating environment revenues
- Sever shipments segmented by the operating environment installed at the factory
- Server revenues segmented by the operating environment installed at the factory
- Desktop/laptop/client operating environment shipments
- Desktop/laptop/client operating environment revenues
- Desktop/laptop/client shipments segmented by the operating environment installed at the factory
- Desktop/laptop/client Server revenues segmented by the operating environment installed at the factory
It's exhausting just thinking about all of the different ways things can be counted. Can you see why a company, such as my former employer IDC, has over 900 analysts? Here are a few questions for loyal readers to contemplate in their spare time:
- How can one accurately and reliably count open source software when it's freely available on the network, people are encouraged to share it with others and suppliers of this software are highly incented to say "oh yes, a gazillion copies of our software were downloaded last week/month/year."
- How can one accurately and reliably relate software shipments to hardware shipments when it is quite possible to install more than one operating system on a system?
- When a company or an individual builds a system out of components, how is that to be counted? How would a research firm even find these lovely systems.
- If a desktop/laptop/client machine is being used as a server, is it counted as a server or as a desktop?
- If a server is being used as a workstation for a single person, is it counted as a desktop/laptop/client or as a server?
Counting angels dancing on the head of a pin starts to look easy doesn't it? (As an aside, I've never understood why these angels would want to dance on a head of a pin. I also fail to understand why anyone would really care what this number is unless they had a contract to manufacture the angel's robes or their dancing shoes.)
What Impact Does Virtualization Have on Market Research?
Each layer of virtualization technology (see Kusnetzky Group Model) can add a new and exciting challenge to the task of counting revenues and shipments. Each layer of virtualization technology has a different impact on the counting process. Some of the layers easily could change the perception of shipments and market share. Let's look into a few of these cases.
- Virtual machine software or operating system partitioning software has a tendency to reduce the number of systems shipped, increase the number of operating environments shipped, increase the cost of that system because it needs to be more heavily configured to do the work. So, operating system shipments and the value of the underlying hardware would increase while the system shipments would decrease.
- High performance computing/Grid computing software has a tendency to increase system shipments, reduce the value of each one of those shipments, increase the number of operating systems shipped and reduce the revenue produced by each operating system shipment.
Which market share figure would be best to consider when looking into these markets? Operating system shipments? Operating system revenues? System shipments? Revenue produced by those system shipments?
Do you think that all of these numbers, each valid in its own way, are confusing to decision-makers? Do you think that journalists are a bit confused by these figures as well?