Most people “love” lists. The psychology of lists has been proven somewhere (no URL provided). On the other hand, analysts love surveys. So I spent a little time digging into the factoids and finer points of portions of a user survey released this week by open source enterprise service bus (ESB) supplier MuleSource. I know from my years at Datapro and IDC that IT users like IT research too. (Perhaps they are normal and like lists as well). For IT staffers, the idea is to see what other IT guys are doing. Unless they are out and out contrarians, like analysts, there is benefit in knowing which technologies peers are betting on.
Take a look at Mulesource’s data yourself. The company gives you its opinion on what the results mean, and I give you mine below. In this world of open source, research and analysis are becoming open source as well. You might also want to look at my opinion on Linux Leveling Off, posted at Research 2.0. It’s based on IDC quarterly data released to the public this week.
A couple of cautions: Only 34% of the respondents were from the U.S., making the mix a little suspect as compared to overall software usage (the U.S. accounts for about 45% of spending, according to publicly released IDC data). Mulesource didn’t publish the survey instrument so I don’t understand some of the tables provided out of context. I also didn’t see any indication of a statistical significance percentage but I suspect it is “wide.” Most important, this is a survey of Mule users so what they say they are doing with Mule cannot in any way be inferred to Sonic, Fiorano, Logicblaze, or other ESB users. But unless Mule ESB users are Vulcans or something, the non-Mule factoids below probably have some universality to them.
According to the release, of Mule ESB users:
• Almost 30% already have their ESB in production; I find this percentage high for any relatively new technology (the ESB in general, not just MuleSource’s) based on my research into the adoption of client/server, ERP, ecommerce, CRM, and so forth over the years.
• About half are “community” oriented, in that they give back custom add-ins to the market; this statistic is probably what is driving the other numbers indicating higher than expected (by me) early adoption. That would be an interesting change in behavior caused by the OSS movement if my theory is correct.
• A major Mule usage is custom integration with legacy apps according to Mulesource’s analysis but I find it hard to reach that conclusion without the survey instrument and without the statistical significance information; clearly is it an important use along with building (presumably new) self-service apps and portals
• Mule users are overwhelmingly JAVA guys, claiming a large adoption of service oriented architecture (SOA) principles
Not coincidentally the group reports big Spring application development framework usage, another possible statistical anomaly.
What I find most useful in such supplier-sponsored and technology-restricted research findings are the factoids separate from the restriction (that is, Mule usage in this case). As I implied above, unless there is some weird correlation between Mule usage and otherwise aberrant behavior, these findings should be more representative of the IT universe as a whole:
• BEA WebLogic, IBM WebSphere and Red Hat JBoss were statistically tied (whatever the precision of the sample) as preferred application servers; Oracle AS trailed badly. I suspect this finding is based on the company not getting a representative mix of database users in their surveying methodology; if they did not have about 66% of their sample running Oracle RDBMS, they did not have a good sample of the IT universe
• 50% are using integration techniques (web services specifically) to connect to customers and suppliers, not just to connect within their own enterprise
• A large percentage of respondents are using MQ Series as transport (although JDBC was leading transport/protocol); in a related question, as many respondents said they were using MSMQ as using TIBCO for actual queuing; Oracle AQ scored well too.
• Following conventional wisdom, there was a high use of Active MQ and JBoss messaging for queuing. But defying conventional wisdom, 60% of these presumably avid OSS fans are working with Windows
Mulesource claims to be the “the leading provider of open source infrastructure and integration software.”
Be careful however. The company uses the now debased JBoss-driven metric of downloads to characterize its market acceptance. I call it the “free AOL CD in the mail” metric. Once JBoss numbers became “public” after its acquisition by Red Hat, it was clear that a vast majority of the millions of JBoss downloads had not been translated into maintenance contracts. In my opinion, such contracts signify market acceptance better than downloads. On the other hand, I could also argue—and have—that the lack of conversions to subscription maintenance simply indicates that the related OSS product really works well.
In related news this week, the 6-month-old Mulesource secured an additional round of financing from venture capitalist Lightspeed. Mulesource will use the investment to double its headcount from 25 to 50, primarily in the area of sales and support.
Leave a comment