Winning and losing with open core

Last week at OSBC I attended an interesting discussion between Matt Asay and Gartner’s Brian Prentice, much of which was spent discussing the relative merits of open core and pure open source business strategies.

Brian has written up his thoughts on the discussion, and I thought I would also add mine since there were a number of areas in which the discussion overlapped with my presentation on business strategies the previous day.

In short, Brian is not a fan of the open core model, despite it being a current favourite of many VC-backed open source specialists. There was a particularly amusing moment when one of the audience members pointed out that Brian’s criticism of open core could upset the sensibilities of most of the vendors in the audience. “Yes?” replied Brian, somewhat bemused, before pointing out that it wasn’t his job to make them feel good about their business strategy.

I think Matt actually summed up the situation with open core pretty well when he described it as “a winning strategy with a lower case ‘w’.” There is no doubt that open core is one of the most popular strategies among open source specialists right now.

As I noted in my presentation, in our 2008 report Open Source is Not a Business Model, we surveyed 114 vendors and found that the open core licensing strategy was used by 24% of them, compared to 27% using a single open source license, and 16% using dual licensing.

For the purposes of my presentation I took another look at those 114 vendors and found that in 2010 30% are now open core, 18% single open source, and just 5% dual licensing (which put some previous discussion into perspective).

However, I also noted that only provided a limited picture of how vendors generated revenue from open source software. In 2008 we were confident that in selecting those 114 “open source vendors” we had identified the vast majority of vendors making money from open source software.

As more and more proprietary software vendors, and software service providers have engaged with open source development, the concept of an “open source vendor” has become meaningless, to the extent that I now think it would be impossible to repeat such a survey.

While open core has been a winning strategy in recent years, I have my doubts about whether it will be the Winning strategy in the long-term. One of the reasons for this, as noted in my presentation, is that it is a product-led strategy (as opposed to pure open source services-led strategies).

There are two trends that are likely to limit the long-term success of open source product-led strategies. The first is that the proprietary vendors are now using open source collaborative development and code to lower the costs and improve the quality of their own product development, and are better quipped to compete with open core vendors thanks to their installed base, larger resources, and product maturity. I have previously argued that open core vendors actually need to become more open to encourage wider adoption and collaboration.

The second is the shift towards software services (SaaS, managed hosting, cloud-based delivery), where the value being delivered is via services, rather than products. While we see some examples of vendors using open core like tactics to encourage SaaS usage by providing bells and whistles not available in the open source version, my feeling is that with software services the value is in the delivery, and that open core approaches become irrelevant (although I do think there is a difference between application- and infrastructure-level projects in this regard).

As we see more workloads shift to SaaS and cloud environments, and we see more open source start-ups born into the “cloud era” we will gain a better perspective on how software services impacts open source-related business strategies. That is one of the main topics we will be considering as we continue our research into open source business strategies.

Tags: , , , , , , , , , , , , , , , ,


#1 Henrik Ingo on 03.25.10 at 4:29 pm

What I don’t get is, why do people even consider open core an open source business model? What is the difference with:

Microsoft taking the BSD TCP/IP stack and putting it into Windows, selling it?
IBM taking Apache software and putting it into Websphere, selling it?
MySQL creating a proprietary monitoring tool and bundling it with MySQL server, selling it?

In all of the above, open source software is involved, but what is being sold is not open source. (MySQL had other business models which where open source, and some of IBM’s might be too, like shipping servers with Linux.)

You kind of make this point by saying that you cannot distuingish open source vendors anymore. The only way you could is to simply reject open core as qualifying as an open source model (which is shouldn’t). So you have something like Red Hat, which really only ships open source software, nothing proprietary. Or you could count MySQL’s dual licensing business, but not the subscriptions.

But if you try to qualify proprietary as open source, then no, it’s hard to know the difference anymore.

#2 Matthew Aslett on 03.26.10 at 3:37 am


If you want to use the term “open source business model” at all then I agree it should not involve proprietary software, so open core is not an “open source business model”. However, what I am talking about is strategies for generating revenue from open source software, in which case open core is in, as is building proprietary software based on open source code. This is why I try not to talk about “open source business models” anymore.

On a side note, I find it interesting that you would consider that shipping servers with Linux might be an open source model – proprietary hardware is okay?


#3 Andrew C. Oliver on 03.25.10 at 4:39 pm

I still do not understand the difference between an “Open Core” vendor and WebSphere. In my mind WebSphere is open core except IBM doesn’t need to brand the constituent parts of WebSphere. The core of WebSphere is by now primarily Apache, Xalan, Xerces, etc etc etc…this can be said of most proprietary software. I do not object to “Open Core” as a strategy. I object to the way you promote it as if it were a form of open source which it is not. Open Core = Proprietary Software with open source used in the product. Windows 7 is “Open Core” because it has open source components. WebSphere, WebLogic, etc. They are all “Open Core”. You have to quibble over “bits” to decide how much “open source ingredients” are required to compose an “Open Core” product. Open Source has a definition. “Open Core” has a marketing campaign (backed by the 451 group among others 😉 ).

#4 Matthew Aslett on 03.26.10 at 3:59 am

Hi Andrew,

Open core has a specific definition which, to me, would not include WebSphere etc. One is building extensions to open source, the other is using open source to build proprietary product, or replace components of a proprietary product. There is a fine line though, and I actually think that by opening up their code to a wider community and focusing on the extensions some open core vendors could evolve to the WebSphere approach (for want of a better phrase). As for whether open core is “a form of open source”. Clearly it is not, but it is a business strategy for generating revenue from open source software (and therefore, arguably part of an “open source business strategy”, as discussed above.


#5 Andrew C. Oliver on 03.27.10 at 1:26 am

And as a strategy, IBM constructs a proprietary product out of a large number of open source components and sells it. “Open Core” companies (aka proprietary software companies with open source marketing) construct a proprietary product out of a large number of open source components (including their own) and sell it. The distinction is very much the stuff of a marketing campaign rather than anything that a mere software developer like me can comprehend. Some vendors are riding our brand (Open Source) to sell proprietary software, pure and simple. And where you do not clearly make that point, you’re committing a classic “bait and switch” or a form of simple fraud.

#6 Matthew Aslett on 03.27.10 at 4:20 am


From development standpoint most (although admittedly not all) open core vendors employ the developers of the vast majority of the OSS code that they use. IBM, in the example above, might employ a lot of the developers of the OSS code they use, but there is a big distinction in terms of the control of the development project and, in all likelihood, the copyright.


#7 Andrew C. Oliver on 03.28.10 at 2:36 am

That is absolutely and categorically false. There is no way that so-called “open core” proprietary software vendors employ the developers of the vast majority of the OSS code that they use. Go look in the lib directory of any of your favorite software. You’ll discover that they use a tremendous amount of open source software and write (by proportion) comparatively little. And that is in a sense the WHOLE POINT of open source. And its totally fine! Its just lying and saying you’re open source when you are not that is NOT FINE. Stealing our message to hawk proprietary wares is NOT FINE. And when we inject these little “controls” that break the efficiency of open source in the pursuit of short term gains while pretending that we’re still open source…it dilutes the meaning of open source and poisons the future for everyone.

Of course if you exercise absolute control over the project, you loose nearly all external contributors…That is simple economics and human self interest.

#8 Matthew Aslett on 03.27.10 at 4:45 am

And if you think vendors are committing fraud and devaluing Open Source, name and shame.

#9 Simon Phipps on 03.27.10 at 6:47 am

OK, I’ll bite. How about:

“Click here to get open source application server WebSphere Application Server Community Edition at no charge
WebSphere Application Server CE streamlines the creation of less critical departmental-level apps on a lower-cost, open source Java EE application server.”

#10 Andrew C. Oliver on 03.28.10 at 3:02 am

fair point simon (below, this software is weird)

#11 Andrew C. Oliver on 03.28.10 at 3:02 am

ok fair point above (but below when i type it)

#12 Open Core Business Model Revisited « James Dixon’s Blog on 03.26.10 at 9:55 am

[…] Matt Aslett : […]

#13 Matthew Aslett on 03.29.10 at 3:36 am


Yes IBM has a track record of confusing statements about WebSphere Application Server Community Edition. I have mentioned this to Savio in the past and they are usually quickly corrected.

That page is highly confusing but, having said that, nowhere on that page does it say “Click here to get open source application server WebSphere Application Server Community Edition”


#14 Matthew Aslett on 03.29.10 at 3:45 am

I don’t know why WordPress does this with comments…

#15 Matthew Aslett on 03.29.10 at 3:50 am


“That is absolutely and categorically false.”

If it is false, how are they able to dual license the software? They either use permissively licensed software or employ the developers so they own the copyright (or both).

“Its just lying and saying you’re open source when you are not that is NOT FINE. Stealing our message to hawk proprietary wares is NOT FINE.”

Who are you talking about? In this whole (almost) year-long discussion we’ve had about this I cannot remember you naming a single vendor that is acting in this way.

If you are accusing companies of lying and stealing and fraud you need more than a straw man.


#16 Reint Jan Holterman on 03.31.10 at 9:49 am

First of all, let me start by saying I truly value your blog a lot! As I’m fairly new to these open source ideas and discussions, I went through a lot of your writings and presentations to grasp a better understanding of what OSS is all about and how companies build business models on top of that. They have been very instructive, thanks!

One thing that strikes me in your current blog post is that you mention that “with software services the value is in the delivery, and that open core approaches become irrelevant”.

How I interpreted this is, ‘no matter how open or closed the software underneath the service, the end user only sees (and pays for) the service rendered’. In other words: the end user does not care about what’s inside the black box on which his service is running.

If that is indeed a correct interpretation, then – in a cloudy world – not just the distinction between open core versus open source would become irrelevant, but also between open source and proprietary (fully closed) source software.

What’s your opinion on this?


#17 Matthew Aslett on 03.31.10 at 10:48 am


I’m thinking of writing a post on this soon, as there is a lot to dive into, but on one level yes the difference between open source and proprietary becomes irrelevant since, as you say, the end user does not care about what’s inside the black box on which his service is running. Except that I think we will see vendors using open source as an on-ramp to the cloud (offering the free, community edition to install on premise with no support, and the supported cloud-based version, as well as services to migrate from one to the other). Of course there is nothing to stop the proprietary vendors also giving the software (if not the source code) for their on-premise software away free as well. Arguably those vendors that are delivering value-added managed services via subscriptions for OSS will find it easier to adapt to this model that those offering proprietary product extensions.


#18 451 CAOS Theory » Sales and marketing for open source – same difference on 04.01.10 at 6:32 am

[…] CTO James Dixon has been defending the open core licensing model against criticism from myself and Gartner’s Brian […]

#19 451 CAOS Theory » Let he who is without proprietary features cast the first stone on 04.08.10 at 10:51 am

[…] cast the first stone Matthew Aslett, April 8, 2010 @ 10:51 am ET If the recent debate about open core licensing has proven one thing, it is that the issue of combining proprietary and open source code continues […]

#20 451 CAOS Theory » While lies beyond the open core debate? on 06.30.10 at 5:32 pm

[…] futile. We have been over the debates about freedom many times before but ultimately, as I recently discussed, it will be market forces that decide whether open core remains a winning strategy in the […]