The rabid factions of the Free and Open Source communities are quick to jump on any report, study, or comment that suggests a proprietary product is in anyway better, cheaper, or even comparable to its open source counterpart (this is particularly true when the proprietary product is from the “Great Satan in Redmond”). They will nit pick every word, every syllable, every hyperlink. It’s really quite impressive… yet, they never seem to get around to giving those studies that support open source products the same level of due diligence. So, that is a role that we at ACT often like to take on… especially with the particularly weak efforts like the European Commission’s FLOSS study and now Alfresco’s recent open source trends study. Reassuringly, we are not the only ones who noticed how flawed Alfresco’s analysis is – Henry Yandell, Director of the Apache Software Foundation, on his blog offered some pointed criticism of “a dumb piece of analysis” in the Alfresco study which “irritated the hell out of
On July 23, open source content management firm Alfresco published a report on a survey of trends in the use of open source software in the enterprise. The report, authored by Chief Marketing Officer Ian Howells, contains some interesting information, especially for software vendors.
Howells’ title (Marketing Officer) really tells us most of the story before we even get to the first page: maximize publicity for their work and their company by drawing conclusions from the survey that are much more far-reaching than the data actually allow. But that is to be expected. The problems arise with the bold claims about trends and survey results sure to grab the attention of the press and the blogosphere – but with no evidence to substantiate these claims.
On page six, the report asserts that open source software is more innovative and provides better value than proprietary software. Howells maintains that, along with the hope to be able to get new technology on the cheap, the driving force behind the adoption of open source software in the Global 2000 is a desire for innovative solutions.
When Howells claims that open source companies are more innovative than their proprietary counterparts, he is propagating a myth of which open source advocates are trying very hard to convince policy-makers, CIOs and others involved in procurement decisions. In effect, he is pursuing the same strategy as the authors of the EU FLOSS study: constantly assert the superiority of open source software and hope that, in spite of the lack of evidence, eventually somebody will believe you.
While it is nice of Mr. Howell to suggest that open source creates the leading edge of innovation, most of the rest of the world looks at Apple as the most innovative company on the planet. In fact, BusinessWeek and the Boston Consulting Group in May 2007 gave Apple the top spot in their The World’s 50 Most Innovative Companies list – for the third year in a row! As Mr. Howells surely knows, even though Apple has open sourced the core of Mac OS X and contributes some code to projects like Apache, Samba and Jabber, it is primarily a proprietary company.
In most places, the report does what it is supposed to do: provide the reader with useful information on the adoption of open source software in the enterprise. On page seven, we learn that a majority of respondents (60.3%) intends to use Alfresco on a Windows platform, showing that Windows “plays a healthy role in the open source ecosystem.”
According to Howells, however, this impression is mistaken. This becomes obvious, he claims, when you question users’ intentions for evaluation and deployment. Of the survey participants, 42% said they intended to evaluate Alfresco on Windows – but only a smaller number (29%) said they want to deploy Alfresco on the Windows platform. From this, Alfresco concludes that “most companies use Linux when they’re serious and want to go into production.”
This conclusion is sure to maximize publicity for the report as “it turns on its head the assumption that users develop and assess on Linux before deploying on proprietary operating systems for peace of mind.” While I admire Alfresco for being so media-savvy (Mr. Howells is the company’s Chief Marketing Officer…), I think it’s problematic to conclude from the report that there is a general trend for enterprises to deploy open source applications on Linux rather than Windows.
There are two issues here. First, the survey participants cannot be considered a representative group of users. The results of the survey show that, in three out of four categories (application servers, databases, and browsers / portals) the respondents strongly preferred open source over proprietary offerings for evaluating and deploying Alfresco. Especially the last category (browsers / portals) gives a good indication of how skewed the sample is: Over 70% of respondents said they use Firefox for accessing Alfresco’s offerings. In the wider world, while the use of Firefox is growing, it does not even come close to having a 70% market share.
The participants of the survey must therefore be considered significantly more pro-open source than the average user. Asking this group if it would like to replace the proprietary operating system in its stack with an open source OS is akin to asking a group of Democrats if it would like to replace George W. Bush with a Democratic President. What do you think the answer would be?
Second, there are plenty of studies out there whose results lead to conclusions different from Alfresco’s – in fact, the alleged trend towards evaluating on Windows and deploying on Linux is “the opposite of what studies of this kind have shown in the past.” Thus, Glyn Moody reported in July that “open source applications are widely used running on top of Windows. Alongside the LAMP stack, there’s […] Windows+AMP (WAMP).” Eclipse executive director Mike Milinkovich states that "[we’ve found that] 87 percent of our downloads are for Windows, and Linux is 9 percent.” He confirms that Eclipse is "very much part of the broader Windows ecosystem.”
Similarly, according to Daniel Chalef of KnowledgeTree, “more than 60% of KnowledgeTree’s approximately 12.000-15.000 monthly open source downloads are our Windows stack installer. One of the primary differentiators of our commercial offering is its tight integration with the Windows desktop and Microsoft Office. Our customers and open source users are running on Windows, like Windows and intend on staying on it.”
Alfresco’s final attempt to attack proprietary software and attract the attention of the press and the blogosphere revolves around its conclusion that the different deployment patterns of Red Hat Linux and Novell SUSE Linux among Alfresco Community members show that customers do not like the terms of the Microsoft-Novell deal.
It is worth pointing out that the polled users did not have to give reasons for their Linux choice. So it’s not quite clear how Mr. Howells can claim that Alfresco Community members’ preference for Red Hat over Novell SUSE in the period from March to May “is not a coincidence” and can be traced back to “customer unhappiness with the Novell-Microsoft deal” – maybe he consulted his crystal ball or read the tea leaves.
There are, in fact, a number of possible reasons why more members deployed Red Hat than Novell SUSE. As Internetnews reporter Sean Michael Kerner points out:
Among the other reasons why Red Hat users may well have increased is the emergence
of a new Red Hat release — Red Hat Enterprise Linux 5 (RHEL) — which was released
in mid-March just a few weeks ahead of when the Alfresco study began. Novell on the
other hand had no similar product launch event at any point near the Alfresco study.
Howells also admitted that the Alfresco study did not ask respondents which version of
RHEL they were using. The study also did not identify any RHEL clone users, such as
CentOS or Oracle Enterprise Linux.
Confronted with these facts, Howells admitted that he lacks the empirical evidence to back up his allegation. “All we’ve got is the raw statistics,” he said, and then conceded that “there may be a number of reasons for [the Red Hat platform growing more rapidly in size than the Novell SUSE platform] but we don’t actually know the specific reason.”
Even if Howells is right and the reason why more Alfresco Community members chose Red Hat than Novell SUSE in the period from March to May is discontent with the Novell-Microsoft deal, from this it cannot be concluded that there is unhappiness with the agreement in the wider world, too. As mentioned above, survey participants said that they would strongly prefer to evaluate and deploy Alfresco on an entirely open source stack. This indicates that the polled users had strong feeling towards open source and felt a close ideological affinity with the open source movement. This is by no means the case for all users of open source software; as Barry Klawans points out in InfoWorld, “for better or worse the business customer […] uses the phrase ‘open source’ to mean many projects” and mostly “cares if [the software in question] solves his problem in a cost-effective manner.” Therefore, if the survey participants do not like the Novell-Microsoft deal, the same feelings can by no means be attributed to the average user.
According to John Mark Walker, “[d]espite the nascent success of open source software” there has been increasing concern among users “about potential pitfalls, such as patent infringement claims.” As Neill McAllister points out, “[c]ertainly, the potential risk to customers, should key software technologies in the products they invest in later be ruled to be infringing, is significant” – customers therefore value indemnification agreements.
In other words, there is a lot of pragmatism among customers when it comes to open source software. Most users see the benefits of business alliances offering them additional protection against infringement claims and couldn’t care less if a few fanatics at the fringes of the open source community allege that alliances of this kind violate the spirit of the movement.
As much as Alfresco would like to make us believe that the pattern observed in the Alfresco Community is repeated in the wider world, this is actually not the case. While it is true that Red Hat’s Linux business is outgrowing Novell’s, this trend was already visible in 2006, before Novell entered into a deal with Microsoft. Thus, according to Nathaniel Martinez, program manager at IDC’s European Enterprise server solutions group, Red Hat had 54 percent of paid server shipments worldwide in 2006, while Novell had 26 percent. This indicates that customers’ preference for Red Hat’s offerings over Novell’s is quite independent of Novell’s agreement with the company from Redmond.
Also, while Novell’s Linux business isn’t growing as fast as Red Hat’s, “overall Linux revenues at Novell have been growing rapidly in the past six months — in the quarter ending in May, its Linux business jumped 83 percent to $19m.” Commentators have pointed out that a lot of this is revenue from the Microsoft deal, indicating that, contrary to what Alfresco claims, the deal has been extraordinarily well received by customers.
Alfresco’s Open Source Barometer contains some interesting numbers. It is a pity that instead of drawing limited, but sound conclusions from the numbers, Alfresco has decided to take some cheap shots at competitors. While its bold claims about open source’s innovative nature and its far-reaching conclusions about users’ sentiments towards the Windows platform and the Novell-Microsoft deal are sure to create some buzz in the blogosphere and the media, they are not backed by facts and thus can ultimately not be taken seriously.