Open Source Mobile: Funambol’s Differences between Community and Carrier Editions
On my way back from the Open Source Think Tank I met my friend Fabrizio Capobianco at his office in Reedwood city, and he took the chance to show me his new MobileWe portal, the Funambol’s version of Apple’s MobileMe.
Funambol business model has always intrigued me, and I asked Fabrizio more about how Funambol manages the difference between the community edition and the carrier edition.
The difference between the Community edition and the Carrier edition is exactly the same, as it was. So we have not changed the business model, is working extremely well: we know that our community is growing fast and we haven’t had any problem with the community like other commercial open source have.
Community edition is for enterprise, while the Carrier edition, that is a commercial product, is for service providers that host our platform for consumers.
That is the distinction we keep, that puts features in one bucket or in the other. So if it is an enterprise that wants to download our server and integrating it with Exchange or Domino and give push email to everyone that hasn’t a Blackberry, that’s the Community edition. If you are a mobile operator and wants to offer mobile mail, calendar sync, pictures sync and all we call MobileWe, then you buy the carrier edition.
The main differences are the portal, the over-the-air configuration of the phones, the scalability to the managed users, the customer support interface.
We do not support commercially the community product, we are not taking any money from community members. If you are a community member and want support, we send you to a partner that can we help make money, doing what you want and give it back to the community.
Funambol keeps layering users and customers depending on their needs. Addressing commercially only the top of the ‘pyramid’, Funambol actually doesn’t upsell its enterprise-based community. Enterprises using Funambol are happy to contribute back their enhancement to get them in the next release, unless they create differentiating features.
Open source core is not bad or good per se, as far as the name confusion is resolved in favor of customers. Funambol makes this difference clear, and consistent. Funambol’s architecture of participation welcomes contributions in code, either in the from of subproject creation by interested third parties, or by coders enjoying the code Sniper bounties.
Even if Funambol retains full control of decision making and IP ownership, there is little or no tension between control and openness. Extrinsic motivations to contribute may vary, but it is a matter of fact that it takes place, like in the Blackberry case.
Mobile operators and carriers are not community users, and it is unlikely that they will ever want to spend time and effort to develop Carrier edition proprietary extensions. Besides that, Funambol – just as every software vendor – has to keep innovating, and speed seems also an enabling factor (think about carriers seeing Apple, Google and Microsoft trying to ‘steal’ their customers!).
When deciding on proprietary features, vendors segmenting their customer base and then the list of features, tend to be prone to errors.
Do System Integrators try to proprietarize your software?
The normal situation is an enterprise that download the software and build something around it. Sometimes S.I. are a kind of bundling our product and sell it as-it-is. In that case we are very aggressive in making sure that AGPL (Funambol’s license of choice) is respected.
How many community users exist?
Recently we did a check to see how many servers we have alive. We have a feature that updates the servers, and checking how many ping we received. It come out that there are about 12,000Â servers. I didn’t have a look at where are they from yet, but I expect to see the same distribution as per the downloads, that last time we checked was 30% USA, 40% Europe, 30% Asia.
It is a pretty impressive number, really, congratulations!
Roberto Galoppini, 
Xavier Giannakopoulos, 
Andrew Lampitt 4:37 pm on March 11, 2009 Permalink
Thank you for your Tweet, Roberto. Just read this blog a moment ago and at first tried to respond by Twitter, then considered email, but then realized might as well respond here.
Wonderful post, Roberto. Yes, I believe Fabrizio Capobianco is the farthest along of anyone I am aware of in the dynamics of commercial open source. Fabrizio is clearly a pioneer of the Open Core Licensing business model as I articulated in August 2008 on my blog. Perhaps he invented it? (I simply coined the term and articulated a basic definition.) I am admittedly fuzzy on the history of some business models and vendors in 2000-2005. At the time of my first open core blog, I did not include AGPL in hopes of being more inclusive, but I added it in my second blog on the topic as I realized the point it addresses is unavoidable. Fabrizio had already been a long proponent of AGPL.
And Funambol’s example brings a good reason as to why copyright ownership is important in Open-Core: Funambol was able to create a unique license to accomodate his correct concerns around GPL v2 (not sure if Funambol used GPL v2 at one time? Anyway, he created the Honest Public License to include a sort of Affero clause) and then switch to AGPL when it came out – license switching is not something you can do without copyright ownerships or some sort of major control. I plan a blog to expand upon the further benefits of copyright ownership.
I believe Fabrizio’s segmentation of features by user type is one great way – and works beautifully in Funambol’s space, but I am not sure it as clear for every other segment as it is for Funambol’s division of community/enterprise and mobile carriers. Consider CRM, RDBMS, ETL, EAI, BI, etc. I would love your and Fabrizio’s thoughts on that – what do you see as the “separate kinds of users” in the corporate world? Seems fuzzy to me. Although it is true, to paraphrase Marten Mickos’ characterization, that some users will always have more time than money (“community) and others will have more money than time (“enterprise”), especially now during this poor economy I think there are more enterprises with limited budgets that would use community editions “for now” and then upgrade later to commercial versions with more features.
I am also curious of your and Fabrizio’s thoughts on the use of visible code in the commercial/carrier license? Is it done? beneficial? Not a requirement?
Ultimately, I agree with a key point of your post:
“Open source core is not bad or good per se, as far as the name confusion is resolved in favor of customers. ”
Yes, the key to any successful model is delivering value to users (both in open source and commercial editions.)
Chris Maresca 9:51 pm on March 11, 2009 Permalink
Re: open core business models. I don’t think that’s anything new, we were proposing those kinds of models to client in 2003. And I don’t think we were the first.
What is becoming clear is that most open source-based companies will have revenue from a variety of sources, each one of which will have it’s own business model. And the mix of revenue streams will be different for each company and it will change over time. Adaptability is the key thing, really.
Chris.
Roberto Galoppini 6:21 pm on March 12, 2009 Permalink
Hi Andrew, ciao Chris.
I like the issues you’re raising Andrew, especially about segmentation.
Talking about OpenOffice.org, just to make an example, I see a (potential) clear distinction between consumers and enterprises.
While individuals basically need a free (as in beer) office suite, medium-to-large enterprises have different needs, included support for packaged-services to migrate macros, etc. Often isn’t just matter of time, at least if you have to manage risks (and in similar migrations you have to, indeed!).
So said, I agree that every situation (package, and the product around it) is different, and how wisely says Chris you need to be flexible, time matters!
Chris Maresca 6:52 pm on March 12, 2009 Permalink
One of the key things that most OSS people don’t realize is that no middle manager in a large organization will risk his job on a piece of open source. Yes, community support may be good, but you still need resources capable of tapping into that. Often, it’s just easier to pay OpenLogic or someone else for that support.
I think this may change in the future as we see more open source being created and supported by end-user companies. It really hasn’t happened yet, at least not at the enterprise level, but I’ve heard rumors.
Chris
Roberto Galoppini 9:53 pm on March 13, 2009 Permalink
I think that Carol Izzo explained it clearly at the think tank CIO panel: mitigating risks is definitely a major issue, no doubt. While in US people might ask Open Logic, here in Europe the one-stop shop is the system integrator, as also Larry made it clear.
So said, while medium-to-large companies are already starting to enjoy open source self-sufficiency, I can hardly see smaller ones going that way.
As usual, customers are always willing to pay for value, but vendors need to deeply understand their core value propositions.
Fabrizio Capobianco 4:59 pm on March 17, 2009 Permalink
Hi Andrew,
thanks for asking my opinion about how can the Funambol model be made general. I actually have a presentation at OSBC about it next week (10:30 am on the 25th), which is going to be very similar to the keynote I gave at the World Computer Congress (linked here: http://www.slideshare.net/fabricapo/world-computer-congress-keynote-presentation).
The gist is: open source for deployment, commercial for hosted. Matches with Marten comment about people that have more time than money (they download and deploy in house), and those that have more money than time (they buy the hosted version). And it fits with the SaaS shift, plus totally solves the separation of communities: they self-select themselves, which is the best because you do not have to do it. No tension. No tight-rope walking.
Happy to discuss it live at OSBC.
Cheers,
fabrizio
Xavier Giannakopoulos 9:59 am on July 1, 2012 Permalink
Hello Roberto,
your article is very positive at the high level, but my personal experience maintaining a Funambol server (free version) is rather rough.
This especially because I have serious difficulties accessing the community handling the “free” version, http://www.forge.funambol.org. I would be interested to understand how Funambol fans out expenses and QoS among the various layers of its products, including which part remains to handle the “free” users.
Cheers, xavier
Roberto Galoppini 8:02 am on July 5, 2012 Permalink
Hi Xavier,
what kind of problems do you have in accessing the free open source version? I believe the code is now hosted at SourceForge, not sure this is your actual problem, though.