Cloud computing: A natural conclusion of open source?

Tim O'Reilly has been stating for years that open source qua licensing is dead, and that the real debate/interest has moved to open source's attributes of open, community-based collaboration. Despite Tim's consistent message, it has only been recently that it has started to sink in for me, with an "ah-ha" moment hitting me halfway through a podcast recording with Geva Perry and James Urquhart on the interplay between open source and cloud computing.

Some of us are just slow, I guess.

Over the past few years, the open-source community, in all its different colors and hues, has demanded that the technology world pay attention to software freedom and its tendency to lower cost, improve interoperability, and more. This freedom is accomplished through open-source licenses like the GNU General Public License.

Along the way, however, open-source businesses started making an adjacent argument that freedom of code ensured maximum choice in selecting a vendor to support that code. While not absolutely true, this argument served to separate the idea of services that complement software from the software itself.

While open-source entrepreneurs initially intended such services to mean "support" and "consulting," the industry has taken open source to one logical conclusion and has crowned "services" as the only important software outcome. In other words, no one cares about Google because it's running PHP or Java or whatever. No one cares about the underlying software at all; at least, its users don't.

Instead, the discussion has moved, as Tim predicted, to the services Google and other new-school "software companies" provide.

Open-source licenses, in this world of cloud computing services, are either irrelevant or obtrusive. Irrelevant, because most OSI-approved open-source licenses don't even apply to network-based software.

Obtrusive, because they focus on the wrong guarantor of freedom. The real questions going forward relate to open standards and open data, because developers don't necessarily need to interact with Google at the source-code level: open APIs, more than open-source code, matter more in networked software.

I believe that open source remains a critical way to backstop the best intentions of those signing up to provide open data. But this should be just one part of the conversation and, as Tim has been telling us, it's not the most important part. ("Architecture trumps licensing every time," he notes.)

What we need, really, is for open source to be married to open standards and open data. Some get this, like the city of Vancouver, which is now moving to an amalgamation of open source, open data, and open standards. Open source is the starting point, but it proves impotent on its own.

The cloud takes open source to its logical conclusion, crowning services (the output of software) as king, rather than fettering us to a discussion of software (the input). Each has a part to play, but open-source licensing is no longer the most important, or interesting, part.

As much as anything, cloud computing has borrowed from open source in terms of its governing principles, which could well be open source's lasting contribution to the cloud. It took a few decades of Microsoft dominance to really get the open-source movement in full swing, but it only took a few months for things like the Open Cloud Consortium to spring to life. Open source has taught us to expect openness by default. The cloud is no different.

Open source, then, has made an indelible imprint on cloud computing. It gave it life by providing the raw material upon which many private and public clouds are built. It gave it a conscience by setting the industry's default principle to openness.

Whether this conscience holds firm is up to us.

Follow me on Twitter @mjasay.

[jrepenning]

We tried "open standards and open data" in the '80s. I'm not convinced that it actually paid out the benefits it sought. Standardized UNIX? Nah ... the standards-waging UN*X vendors flamed out, and the Open Source folks had to rescue us all. Standardized networking? TCP/IP, you bet. OSI? Not so much.

The Cloud revolution/fad surely requires open protocols and data, no question. Some other things as well, like open platforms (that is: as an application developer, I don't want to be locked in to a single cloud vendor). But that's all about knocking off the rough edges of the new idea. It won't make "the cloud" into the fountain of creativity that created the hallmark open source products. It's portability, not freedom.

[zyxxy]

Ethernet is also a standard. As is PCI (and by extension, PCI-e) WiFi is a standard, and so is BlueTooth.

So yes, open and published standards are good. That includes open document formats. That includes published XML interfaces into networked services.

[vikinzer]

I'm sorry Matt, but I have to almost completely disagree with you. While open standards are important, and we do need to worry about them I agree with Richard Stallman on the fact that cloud computing where you purchase services puts WAY too much power in the hands of a corporation like Google. Cloud computing infrastructures more similar to Drupal are built on open source and are necessary to achieve true openness. In a completely open world, where freedom is guaranteed I would have the freedom to create my own cloud, that I could access from anywhere and achieve all the computing advantages of cloud computing for my organization, without putting the power in a large corporate shepherd, who with a bad management shift begin to abuse their power. I've said it in comments on your posts before, power corrupts and total power corrupts totally. The power of open source is that it decentralizes that control, and creates a situation where the company has to provide a value proposition, because if they don't someone else will. Open APIs are no more than the open standards Microsoft used to create their Monopoly, and once they had the Monopoly surprisingly things became a bit less standard.

I want software like Drupal, where I can deploy it and then use the cloud on my own. I know Drupal isn't an ideal example, but I use it as cloud computing. I have installs of it where I go in them and compile recipes and notes for projects I'm working on, and create collaborative systems, all without sacrificing control of my data. That is not to say that I don't think people should use corporate cloud computing services, but I do think it is important to know that at any time someone could pull out and create a cloud for their own use. It keeps the corporations honest, and creates the exact same power balance that makes open source client side software so important for user freedom now.

Matt, I know you respond to comments sometimes, and on this one I am very interested in your thoughts on these observations. So if I could talk you into a response in this comment thread I would really like to read it.

[tim.hawkins_dotmac]

I have to agree with vikinzer, you have to have an vendor interchangeable source for all services you consume, I recently went through the pain of having to recode an application because a public api provider changed the TOS of an api we where consuming. The spec of the API did not change, but its conditions or provision changed. With open source if a license changes you just freeze at the level where the license was acceptable until you make other arrangements. If an online API provide changes the Terms you are in trouble, there is no old version to retain.

It is noticeable that the vast majority of cloud providers have failed to sign up to any standardization efforts, they dont want inter-changablity, because it breaks lock-in.

Im a heavy Amazon EC2 user, currently for production use EC2 is only available from one supplier, ( although there is an opensource version, check out http://www.eucalyptus.com/ ) and if amazon suddenly decides to triple thier prices, then im stuck, i have too much committed to their infrastructure.

Yesterdays Google outage also shows what happens when a key supplier of services suddenly stops providing dial-tone, and there is no practical way to back up these services, the effort involved in developing and provisioning alternatives makes the benefits of using the cloud originally difficult to justify.

[ArtInvent]

Completely agree with vikinzer and tim.hawkins. I have very little interest in an IT world where the cloud has enabled large monolithic corporations to lock me in to services as Google is fast on the road to doing. I think it's going to be very important to people interested in choice and freedom and innovation in IT to ensure that moving into the cloud is not used as an opportunity to bury all that's been accomplished by the open source movement. that's exactly what I think will happen if cloud computing is not fundamentally open in every sense, and that certainly includes the licensing of both the client and server software, and certainly not just the API's. As has been pointed out, MS and Apple both have pretty free and open API's for a lot of there stuff - that's simply a prerequisite for attracting developers to what are very proprietary and restricted platforms.

But this vigilance is going to require us to see open source as more than just a means to a particular end in business.

I would also add that consigning open source licenses to the 'irrelevant' heap is incredibly myopic. That's probably because this blog focuses on open source as a software business strategy pretty much exclusively, as if it were just any other tool in the business world. But open source is at it's root a much broader and deeper phenomenon than that. It's a philosophy and a social movement. It's given rise to other 'open' movements like open design, open hardware, wiki, copyleft, CreativeCommons etc. that actually have nothing to do with software at all, and in many cases nothing to do with IT whatsoever. Business only has taken it up as a strategy to fight the old entrenched closed companies that have strangled business opportunity for newcomers.

[Matt Asay]

I do think that code is the guarantor of other freedoms, and so I think we agree there. I just don't think open-source code is enough. (And I do find it a bit cheeky (not on your part, but on RMS') to hear about problems with the Web borrowing open-source code without giving back, since RMS or, at least, the FSF, explicitly chose not to include notice of hosted distribution in GPLv3. AGPL was an add-on kluge.)

I suspect we'll see the open clouds you describe as a response to the currently closed clouds. But as I said: it's telling that we're so early in the cloud conversation and already talking openness. That's a result of all the good work open-source advocates have done.

[Matt Asay]

I do think that code is the guarantor of other freedoms, and so I think we agree there. I just don't think open-source code is enough. (And I do find it a bit cheeky (not on your part, but on RMS') to hear about problems with the Web borrowing open-source code without giving back, since RMS or, at least, the FSF, explicitly chose not to include notice of hosted distribution in GPLv3. AGPL was an add-on kluge.)

I suspect we'll see the open clouds you describe as a response to the currently closed clouds. But as I said: it's telling that we're so early in the cloud conversation and already talking openness. That's a result of all the good work open-source advocates have done.

[throktar]

Cloud Computing is nothing new. Just a rehash of the "mainframe" era. Open Source as a viable business model will ultimately fail. Time to move on...

[tochirag]

There are strong synergies between Open Source Software (OSS) and cloud computing. The cloud makes it a great platform on which OSS business models ranging from powering the cloud to offer OSS as SaaS can flourish. There are many issues around licenses and IP indemnification and discussion around commercial open source software strategy to support progressive OSS business models. I do see the cloud computing as a catalyst in innovating OSS business models.

I have a detailed post on my blog:

http://cloudcomputing.blogspot.com/2009/01/open-source-software-business-models-on.html

[pentest]

Try posting that again with fewer buzzwords, it might make sense then.

[vikinzer]

WOW! pentest hit this nail on the head. I've never seen to many big words say so very little before in my life.

[Matt Asay]

I, too, worry about cloud lock-in. The problem is we're already sprinting pell-mell into this state, dumping our data into Google, Amazon, etc. So far, we haven't seen a big open rival to these closed clouds. Is this because consumers just don't care? At present, I do think that's the case: people just want functionality, and aren't thinking about the longer-term issues like *real* control of one's data. But I suspect that will come as clouds fail, taking data with them.

[vikinzer]

I have to say it's good to hear you say that. I agree that open source isn't enough, but one thing I think we saw early on is that individuals who push open source tend to also push open standards. The whole culture of "choice" is ironically enough built on a lack of choice in data formats. You kind of need to know everyone is going to interface properly with the same data in order to have control over your software and interface. So if you want to change things up you know everything will still work.

I think that it will be very telling to see where Ubuntu One ends up going. Canonical taking a proprietary approach to their server side offering is making some major waves, which will prompt a lot more conversation on this topic. It may also be that they prompt a lot of conversation and then end up releasing the code later like they did with launchpad. I personally have no problem with a company wanting their software offering to reach a certain level of polish before releasing them. We all know our community often forgets what beta, alpha and proof of concept means because we're so used to running beta software day in and day out as general use. (Google's fault much?) Either way, it's going to be an interesting ride.

[jrepenning]

Still not clear to me what you're claiming about the relationship between OSS and Cloud. I think you demonstrate that the OSS experience has set expectations that the Cloud now ought to meet. The same can be said of government, and a number of other areas as well.

But calling cloud the "conclusion" of open source appears to imply that you think there will be no more open source software. Was that intended?

Maybe I'm touchy on that one: I'm worried that "open source" is sliding into "commercial open source," which erodes into "glass-house development," ending in "faux-open source," an appearance of openness without its substance. Was that the "conclusion" you had in mind?

[semanticloud]

The big question is... what happens when cloud computing interacts with linked data.. how do things like Nepomuk factor in this whole equation?