It's been nearly two and a half years since Oracle announced their Oracle Unbreakable Linux program. Flash forward to present day, Oracle is still facing a strong perception in the community and tech media that "Unbreakable's" mission is to make new Oracle customers solely at the expense of Red Hat.
A perception Oracle really wants to change.
Set the Wayback Machine for... 2006
There's a lot of history between Oracle and Red Hat, to be sure. The two companies were solid partners in the first half of the decade. Oracle's favored status as a Red Hat vendor meant that its products worked very well with Red Hat's enterprise products. At the time, there were some complaints in the Linux community that the products worked too well, and allegations were made that there were a lot of unreleased unique hooks just for Oracle into Red Hat's APIs that no other Linux distro had.
That argument became moot when the balance was altered in 2006. In the early part of that year, various reports indicated that Oracle was looking to purchase middleware vendor JBoss. In April of that year, Red Hat acquired JBoss instead, thus putting Red Hat on a competitive stance with Oracle.
Winding up to the fall of that same year, there was much speculation on what Oracle would do next, since it seemed clear that their partnership with Red Hat was effectively over. Rumors that Oracle would release a distribution based on the hot rising-star Ubuntu distro, or partner with Ubuntu's corporate vendor Canonical, were widespread at the time.
To the surprise to most observers, Oracle instead announced the Unbreakable program in October 2006, which provides support to existing Red Hat Enterprise Linux (RHEL) customers as well as deliver Oracle Enterprise Linux (OEL) to non-Linux customers. OEL, it was also revealed, would be a bit-for-bit copy of RHEL, with only branding changes made within the product.
The move was initially viewed by the Linux community as an attack to Red Hat, though business analysts applauded the move. In 2005, Red Hat was reported to hold 61 percent of the commercial Linux market. Oracle's CEO Larry Ellison took the opportunity to talk up his company's new strategy, stating that OEL would surpass RHEL, Oracle's support plan would cost less than half of Red Hat's plan.
While all of this would seem like so much corporate positioning, several reports at the time came out with information that was misreported, including the notion that Oracle would offer customers its own bug fixes to the Linux kernel and not release those fixes back upstream to the mainline Linux kernel.
Red Hat's own reaction was one of derision, as the Raleigh-based company launched an "Unfakeable" Linux spoof campaign, setting itself up as the genuine article and indicating "Oracle has stated they will make changes to the code independently of Red Hat. As a result these changes will not be tested during Red Hat's hardware testing and certification process, and may cause unexpected behavior. Hence Red Hat hardware certifications would be invalidated."
Talk of a "fork" exploded into the fore, with many in the community wondering if Linux was about to go the way of the UNIX fragmentations of the 1980s. To date, however, Oracle has consistently maintained OEL as a RHEL clone, and no such forking has taken place.
But even today, the bad blood between Red Hat, the community, and Oracle still exists.
Oracle's Quiet Outreach
Instead of forking the Linux kernel, Oracle has actually made significant contributions to the mainline Linux kernel development, according to Wim Coekaerts, VP of Linux Engineering. One of the biggest contributions to date is the code for the next-generation filesystem BtrFS (often pronounced "butter FS").
In a presentation to the Linux Foundation's End User Summit last October, BtrFS founder and Oracle's Director of Linux Kernel Engineering Chris Mason, outlined the cutting-edge capabilities of the new filesystem, including better integrity checksums dynamic inode allocation, and online defrag and filesystem checking.
Mason indicated that even in the early stages of BtrFS development, some of the preliminary code has been integrated into the mainline kernel, which Coekaerts confirmed in an interview this week. BtrFS is widely regarded in the community as the heir apparent to the ext3/ext4 filesystems.
Coekaerts also highlighted Oracle's participation in the Data Integrity Initiative (DII), an effort started in 2007 with Emulex, LSI, and Seagate to provide standards-based, end-to-end data integrity for databases on enterprise storage systems. DII's first set of code was accepted into the Linux kernel in late 2008, reducing data corruption by decreasing the potential for incorrect data to be written to disk, and decreasing application and database errors.
Coekaerts described the project as an effort to catch errors in data at the write phase of data storage. These write-failure catches would be much more efficient than catching failures at the read phase, since data might not be read for months. At that point, Coekaerts explained, any discovered errors would mean having to go back to backup to restore the data.
So if Oracle has been so active in Linux technology development, why is it still vilified in the Linux community? Coekaerts sees some of the problem in the layered aspects of the community itself. Kernel maintainers such as Andrew Morton, Linus Torvalds, and Greg Kroah-Hartmann, for instance, see the value of Oracle's contributions, because "they know what we do."
The next layer up, the general Linux community, is not as familiar with Oracle's work in the kernel and storage spaces, and the animosity generated in the company's break-up with Red Hat still colors opinions of Oracle. This is particularly frustrating to Coekaerts, since the Unbreakable program has hundreds of satisfied customers and, by virtue of Oracle's own internal deployment of some 40,000-50,000 Linux systems, "Oracle may be one of the world's largest users of Linux."
The Oracle-Red Hat Dance
Coekaerts stressed throughout the conversation that "Oracle is not out to kill Red Hat." Instead, they are trying to offer a better support plan than any in the industry, and give customers new to Linux a solid platform upon which to build customer's Oracle stacks.
Still, a direct conflict is not a hard conclusion to which observers can jump. Many of Oracle's own marketing statements have directly compared it's Unbreakable offerings to Red Hat's.
"It's fair to compare us with Red Hat, because we're trying to do the same things they are, only better," Coekaerts stated.
So why not separate itself from Red Hat even more by offering its own unique distribution? Coekaerts again dismissed this idea as he has in the past. Adding another Linux distro to the market would be counter-productive and, more importantly, too expensive for independent software vendors (ISVs) to develop apps for yet-another distro, Coekaerts explained. Even compliance with the LSB standard would not help, he added, since much of what a separate OEL would need would be at lower level than the current standard manages.
Going in the other direction, by just offering a global support solution with no software, would not accomplish much, either. Oracle believes there needs to be a Linux platform to offer their customers, and RHEL's pricing prevents that.
"We're offering [OEL] only because Red Hat requires customers to pay for binary downloads of [RHEL]," Coekaerts said. "If RHEL was free, we would have never done OEL."
Oracle, for now, seems invariably tied to Red Hat, and continues its dance with its partner-turned-rival.
But Oracle has had a couple of years to learn the steps to be a contributing member of the Linux community. There are signs that Oracle is about to try more solos, positioning itself on its own Linux technical and business merits instead of being endlessly compared to Red Hat.
Based on the contributions they have made so far, a stand-alone performance should bring real benefit to Linux.
