Policy —

Red Hat looks out for OSS community with patent settlement

Red Hat has settled a patent dispute with Firestar and DataTern over object- …

Red Hat announced today that it has settled a patent dispute with Firestar and DataTern over two patents, including one that covers object-relational database mapping—a technique that is used in Hibernate, a component of the JBoss middleware stack. Red Hat says that the terms of the settlement will offer broad protection for upstream developers, all derivatives, and downstream distributors, as well as Red Hat's customers.

JBoss, which makes an open source Java application server, was acquired by Red Hat in 2006 for $350 million and currently operates as a division of the Linux distributor. Shortly after the completion of the acquisition, Firestar filed a patent suit against Red Hat in the patent-friendly Eastern District of Texas, claiming that Hibernate infringes on US patent #6101502. Firestar later assigned the patent to its partner DataTern, which also joined the lawsuit. Critics say that the broadly-worded patent potentially covers the entire concept of object-relational mapping.

The financial terms of the settlement have not been disclosed to the public, but Red Hat says that it will protect practically everyone who uses the allegedly infringing code—unlike the controversial patent agreement between Microsoft and Novell, which only guarantees protection for Novell's customers.

"Typically when a company settles a patent lawsuit, it focuses on getting safety for itself," said Red Hat vice president and assistant general counsel Rob Tiller in a statement. "But that was not enough for us. We wanted broad provisions that covered our customers, who place trust in us, and the open source community, whose considerable efforts benefit our business."

Although this deal will end all pending litigation against Red Hat by Firestar and DataTern, the company still faces a lawsuit from patent-holding company IP Innovation over a patent on a "user interface with multiple workspaces for sharing display system objects," which was originally filed by Xerox. The same patent has also been basis for similar lawsuits against Apple and Novell.

Red Hat, which emphatically states that "software patents impede innovation in the software industry and conflict with open source development and distribution," clearly went to great lengths to ensure that the terms of the agreement would conform with open source licensing obligations and serve the interests of the broader open source community. The positive resolution of the litigation with Firestar and DataTern will likely be welcomed by the Hibernate developer community, but in some ways it might also represents a missed opportunity.

A close look at the patent leaves me with the impression that it is very likely not valid, and the fact that Firestar and DataTern were willing to agree to such permissive settlement terms rather than following through with litigation is a pretty good indication that they were looking for a settlement from the start and don't necessarily have a lot of faith in the patent's chances in court.

If Red Hat had followed through and gotten the patent invalidated rather than merely settling, it could have prevented future litigation against other open source projects that use similar technology. Red Hat probably decided that the settlement it finally reached in this case offered the best compromise in that it incurred less cost and risk than fighting for invalidation while also still providing adequate protection for Hibernate and Hibernate's community.

Some software freedom advocates are celebrating this patent agreement because they believe that it demonstrates a GPL-compatible approach to patent licensing. Indeed, it does, but an important point to consider is that the strength of a patent dictates the amount of leverage that a patent-holder has in negotiations. If the case had involved a much stronger patent that couldn't be invalidated, then the patent-holder would almost definitely never have agreed to such permissive terms. This kind of patent agreement is the best possible outcome short of invalidation, but it isn't going to be consistently achievable.

Channel Ars Technica