3 Comments

Summary:

Balancing an open-source community with commercial interests can be difficult, which is why HPCC Systems sought the help of Bruce Perens before open-sourcing its eponymous big-data-processing software. Essentially, the company either ensures the existence of a free version or pulls contributed code.

copyright2

Balancing an open-source community with commercial interests can be difficult, which is why HPCC Systems sought the help of Bruce Perens before open-sourcing the code for its eponymous big-data-processing software. I covered the open-source news last week. Afterward, open-source pioneer Perens directed me to an essay he wrote on the HPCC Systems site explaining the new licensing model he helped create for the software that aims to disrupt Hadoop’s big data dominance.

When HPCC Systems launched in June, LexisNexis CTO Armando Escalante told me the company thought it had to open-source its code in order to compete with Hadoop, an Apache Software Foundation project, for developer mind share. HPCC might not have too much of an issue selling large enterprise on its software, but Escalante acknowledged the importance of keeping the product relevant by engaging a community of developers that will keep its features up-to-date and valuable for new use cases.

But HPCC isn’t a traditional open-source project. It has been a proprietary product within LexisNexis for years and is already serving major customers. As Perens explains in his essay:

For this exact reason, the decision to release the HPCC platform as Open Source needs to be justifiable through a sound business plan. The goals of this plan are to capitalize on the innovation and new ideas that come with an Open Source community, develop new use cases from the Open Source users, continue and increase the relevancy of the product by keeping up with Open Source development, and collect direct revenue from the product — not just revenue from ancillary products like support and training.

Dual licensing, in which products have both open-source and commercial licenses, has been the standard way of managing this, but that model comes with its own issues. As Perens explains in great detail, a primary issue is around copyright. The problem essentially is that companies require copyright assignments (e.g., the MySQL contributor agreement) from contributors so that the company may put contributions under a commercial license.

The fear with this approach is that a project owner will make all future releases proprietary, taking developers’ intellectual property contributions — or at least their inclusion in a commercially supported free product — with them. This can lead to a lack of contributions around the core code base, requiring companies to undertake much of that development themselves while contributors focus on extensions, forks, embedded versions and other related efforts that don’t directly enhance the core project.

The covenant

Bruce Perens

With that in mind, HPCC Systems and Perens developed a new method of dual licensing, which Perens calls the covenant. Essentially, it represents a quid pro quo between the contributors in which companies can get the copyright rights they need to develop products while still giving developers an incentive to contribute.

In the HPCC Systems contributor agreement, the company gets contributors’ copyrights, but it pledges to continue developing and supporting an open-source version of the product for three years after each new code contribution it accepts. It’s a rolling agreement, similar to how every change in your wireless service or each phone upgrade locks you into a new two-year contract.

If HPCC Systems wants to discontinue the open-source version before three years are up, it must donate the existing code to a nonprofit organization that will support the code under a permissive license. If it doesn’t, HPCC Systems must remove the disputed code from its product altogether. It’s an interesting approach that arguably requires both parties — the company and the contributor — to put skin in the game.

However, one downfall for developers is that, presumably, they’ll have to pursue litigation to get this remedy enforced. That could be expensive and time-consuming, especially if their individual contributions and contracts are construed as unique transactions or sets of facts that don’t warrant the joining of plaintiffs into a single lawsuit that would allow them to split the costs. The hope has to be that HPCC Systems wouldn’t want to get involved with this potentially messy litigation, either, and that it will honor the terms of the contract.

What do you think? Is HPCC Systems’ covenant the future of open-source dual licensing, or is there still work to be done?

Feature image courtesy of Flickr user MikeBlogs. Bruce Perens image courtesy of Flickr user Friprog

  1. Hi Derrick,

    I am assigned to work on objections to the Covenant. Many commenters pointed out that the KDE Qt foundation agrement provides a mechanism to activate the permissive license release to the non-profit without a lawsuit, and lists a more concrete set of conditions. Obviously I will be discussing this with lawyers, and I’ll get back to you if something comes of that.

    Share
  2. Vladimir Rodionov Wednesday, September 14, 2011

    This is a strategical mistake which can cost LexisNexis its future. Without any doubts, HPCC will loose its battle in open source community to Hadoop. They won’t get any significant traction or interest because :

    1) its C++, I presume. The language is not so popular among open source developers whose primary education revolved around simple vanilla OOP languages like Java.
    2. Its C++ , hence far less flexible than JVM based system can offer.
    3. Too late to be first.
    Open source development model has its advantages and disadvantages as well. The best strategy for LN would be to exploit OSS disadvantages and offer to the market what other OSS projects can not give:

    1. Solid qualitative product.
    2. Predictable and regular releases.
    3. Extensive feature set.
    4. Excellent customer support.

    You do not have to give out your product for free to compete with OSS.

    Share
  3. George Hernando Sunday, November 6, 2011

    I agree with Vladimir’s comments.
    From a developer perspective looking at HPCC as a possible Big Data tool, the first two things that stand out as problems are:

    - GNU Open Source license is not friendly to any commercial application development. An Apache-like license, like that used by Hadoop, is much preferable.
    - Extendability is done using C++, rather than Java or scripting language like Javascript

    Share

Comments have been disabled for this post