eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
The Linux 2.6 production kernel promises to be the most advanced open-source platform developed to date, according to computer scientists who have been putting the 2.5 development kernel through its paces.
The 2.6 kernel, expected to be released by late next month, will move Linux further into the enterprise, though it will still have a ways to go to meet the demands of the largest enterprise database applications, said Tim Witham (pictured left), lab director for Open Source Development Lab Inc., here last week.
“Right now, its an absolute no-brainer to use Linux in any of the infrastructure-type services and smaller databases,” said Witham. “But the 2.6 kernel will move it up to bigger database applications. Thats the big one and the next real jump for Linux.”
Specific features driving the forthcoming 2.6 kernel toward database services include an enhanced scheduler and a threading library, Witham added.
OSDL was established in 2001 as a global consortium dedicated to accelerating the adoption of Linux in enterprise computing.
“If an enterprise is today running 24-way SMP [symmetric multiprocessing] and its large database, with failover for all its components, then, no, Linux cant do all that now,” said Dan Frye (pictured), director of IBMs Linux Technology Center, also based here. “But in the future, it will be able to. But as very few customers run just that, they deploy Linux on other workloads.”
The 2.6 kernel, which could make its way into commercial distributions in three to six months or less following its release, will support large amounts of memory and a large number of threads, Witham said.
It will also offer improved networking performance and increased storage and types of storage—all the things needed for databases and better performance, Witham said.
Frye said that IBM has conducted several multiweek tests of the 2.5 development kernel on a number of eight-way SMP systems under its Linux Test Project, a set of several thousand test cases and test suites against which the code was run.
Page Two
The 2.5 development kernel will become the production 2.6 kernel once it is stable enough, the code is frozen and enough bugs have been fixed, Frye said.
“There are still several hundred defects open,” Frye said. “There are some issues and things that need to be improved before it becomes production 2.6.”
The timing of that transition will depend on Linus Torvalds, the creator and chief developer of Linux. But the improved stability of the 2.5 kernel, even at this stage, could mean that Linux vendors will be able to release distributions based on 2.6 more quickly.
“My expectation is that Linus wont move to 2.6 until things are better than they are today,” Frye said.
Torvalds has told developers working on 2.6 that he hopes to have the 2.7 kernel opened up by the Kernel Summit at the end of July.
According to Witham, a number of features that the development community believes are not yet ready to be incorporated into the 2.6 kernel may very well be pushed to 2.7.
Among these features are support for complete Non-Uniform Memory Access as well as an EVMS (Enterprise Volume Management System), which deals with the difficult and controversial issue of volume management, Frye said, adding that 2.6 would be better than 2.4 in terms of volume management even without the EVMS.
Oracle Corp. and Red Hat Inc. officials have also previously called for volume management. Wim Coekaerts, principal member of Oracles technical staff, in Redwood Shores, Calif., said: “We would like Linux to have a Logical Volume Manager. The 2.6 kernel will have a device manager, but we need an LVM.”
Paul Cornier, executive vice president of Red Hat, in Raleigh, N.C., agreed.
“Making a more generic cluster file system is important to us, as is an industrial-strength Logical Volume Manager,” Cornier said. “A distributed lock manager completes things. This is functionality that needs to go into the operating system but is unlikely to be found in the next [kernel] upgrade.”
IBMs Frye said that theres clearly a need for an improved volume management system and that Linux is not yet good enough in that regard.
Most Recent Linux Stories:
Search for more stories by Peter Galli.