Author: Jay Lyman
Kusnetzky understands the frustration of friends such as HP’s Linux Vice President Martin Fink, who complained at the recent Jboss developer event that open source software is not adequately covered by
Kusnetzky — who was nice enough to talk with us about how industry is sliced up by analysts to gauge shipments, revenue, growth and more — indicated that Fink has reason to be frustrated. However, he defended analysis of the IT industry in general and how it is adapting to change, and suggested that if Fink
wants to find the information he believes is lacking, he would probably have to convince his company to fund a study on it.
While such studies are not infallible, they often provide vendors, venture capitalists, and others with insights that help them gain, take or keep market share.
When the Open Source Development Labs (OSDL) was wondering about a discrepancy between software and hardware data on Linux deployment last year, the nonprofit’s CEO, Stuart Cohen, went to IDC with the dilemna. Kusnetzky said that while
IDC’s software study showed Linux accounted for about 20 percent of mid-level servers, IDC’s hardware data showed Linux running on 12 to 17 percent of the same types of systems.
“OSDL likes the research we’ve been doing, but they saw the fact that the software numbers and hardware numbers could lead one to conclude there was disagreement among our different analyses,” Kusnetzky said. “There was not disagreement. They were looking at different things.
“OSDL wanted a more complete picture of non-traditional Linux deployment,” he said, adding that while about 80 analysts spread across various software categories were enough to accurately cover 90 percent of the market, OSDL was looking for a fleshing-out of the other 10 percent. The non-traditional category turned out to be significant, joining secondary operating systems, partitioned
operating systems, guest operating systems, dual-boot systems, etc., and bumping Linux up to a “mainstream solution,” according to Kusnetzky’s colleague Vernon Turner, IDC group vice president and general manager of enterprise computing.
released last December and described by IDC system software Research Director Al Gillen as a “more holistic view for the many different manifestations that Linux can take on,” showed an even broader market for shipments and redeployments of Linux, which increased by 36 percent
over the net new shipments previously measured in 2004. The study also found fast growth in packaged applications and infrastructure software running Linux, a market expected to top $14 billion less than four years from now because of an anticipated compound annual growth rate of more than 44 percent between 2003 and 2008.
“OSDL wanted to come up with a number,” Kusnetzky said, referring to a “total” Linux market forecast to grow beyond $35 billion by 2008. “They wanted to be able to tell their sponsors what is the real size of the Linux market.”
Kusnetzky explained that the non-standard study involved standard IDC survey information from more than 2,400 companies and 15-20 suppliers centered on hardware and the enterprise software stack. He said such surveys, which typically ask the same questions when updated so history and trends can be projected — serve as “snapshots
in time,” a term that OSDL’s Cohen has also used to describe the study.
“It allows a snapshot in time — not just as hardware shipments, but also free downloads, and replicas, and re-provisioning. That was the whole idea of that report,” Kusnetzky said, adding that no one had asked for such a study before.
Cohen told NewsForge last year that OSDL paid for the study to establish a baseline, but said OSDL was not planning sponsorship of another, similar study, hoping instead that IDC would take up the task on its own.
Last year’s study utilized existing IDC research on server and PC hardware and software, as well as new
research, including a demand-side study covering customer adoption, plans, and perceptions relating to Linux in 10 countries around the world.
Kusnetzky, who said at a conference in February that a follow-up to the OSDL-sponsored Linux market study of 2004 was not budgeted for ’05, reiterated recently that his firm was unlikely to repeat the study unless a sponsor paid for the work.
“It may or may not happen,” he said. “It depends whether new subscribers ask us to update that snapshot. A lot of times, we focus on what our customers are asking us to do for them.”
Open source, closed lips
Dedication to IDC customer demand aside, Kusnetzky seemed sincere in his sympathy for executives such as Fink, who complained that software licenses are no way to measure software which can be infinitely duplicated and freely
distributed throughout an organization.
Kusnetzky said Fink is not the only tech executive who would like to peer deeper into the makeup of the Linux and open source market.
“Marten Mickos wants to go to another source and say, ‘How much Linux use is out there, what machines is it being used on, how did it get there, what did they do with it?'” Kusnetzky said, using the MySQL CEO as an example. “Some information is relatively easy to get and we publish it. Some information is difficult to get, and we don’t
do it as standard. Some, we don’t know how to get it.”
Kusnetzky indicated that there is a certain degree of Linux deployment, for example, that cannot be measured because it is kept quiet by developers, IT managers, or executives who — mostly for policy and bureaucracy reasons — do not want to reveal their open source use.
“There is one part of open source usage where people don’t want to be discovered,” he said. “It doesn’t matter how well you do the survey, you wouldn’t find it.”
Kusnetzky said an individual or project team may have Linux deployed internally, but will not reveal it because of company software usage and licensing rules or just perception. At the same time, IT shops are under tremendous pressure from company management to deliver results while cutting costs, and management may not ask or even want to know whether a worker or group is using Linux, as long
as it is producing.
Reiterating his sympathy for a frustrated Fink, Kusnetzky said the business leader probably wants the same information he sees for Windows deployments made available for Linux. However, he added that such a level of granularity is not always possible, particularly if survey respondents are keeping quiet about Linux and unconventional deployment methods are being used.
“Open source software is much harder to track,” Kusnetzky said. “That makes [Fink’s] job much more difficult. I would suggest he look at what we did with OSDL.”
Kusnetzky noted that IDC is working to adapt to changes in the IT industry, and is striving to bring the same level of detailed research it does on hardware, services, printers and other subjects to Linux and open source software. He agreed with Fink’s contention that software licenses are becoming a less and less useful way to measure software deployment given the advent of multi-core processors, virtualization,
and of course, Linux and open source, adding that the issue is a hurdle for both analysts and users.
“That is one of those things — software licenses — that in IDC’s view is one of the significant challenges of the next five years,” he said. “We understand the compelling need for vendors to see revenue. We also understand organizations are looking to cut costs, and they
want license agreements to reflect actual usage. There are shades of gray, and what is equitable for operating systems might not be equitable for database software.”
Nevertheless, Kusnetzky said that the endless search for ways to cut costs presents a big opportunity for Linux and open source — particularly for those in the Linux server market, which is poised to grow nearly 25 percent a year through 2008, when IDC believes the market will top $11 billion.
“Organizations are looking for cost savings, almost to the point of hurting themselves,” Kusnetzky said. “Open source really has to focus on the true savings they have to show and how what they’re using allows organizations to function effectively.”