Linux 2.5 Due In 2002

The next version of the Linux kernel, due sometime in mid-to-late 2002, will offer improved support for enterprise, mobile and Internet use, said Ted Ts'o, Linux kernel developer and principal engineer at VA Linux Systems.

During an interview at last month's first-ever Linux Kernel Development Summit, Ts'o described a host of proposed new features for the forthcoming Linux 2.5 kernel that could make the operating system a more powerful competitor against Microsoft Windows 2000 on servers and laptops.

"There's not a lot of desktop [enhancements]; there's a lot more focus on the server space," said Ts'o, adding that a new networking device interface, improved networking performance and asynchronous I/O support will elevate Linux's stature in the corporate world.

Improvements include boosting scalability beyond the two terabyte device limit; support for asynchronous I/O, which will speed up the enterprise database performance; support for new IETF-approved Internet networking protocols; and enhanced support for Intel's Advanced Configuration and Power Management (ACPI) standard, Ts'o said. Linux advocates aim to support the hibernate and sleep mode of the ACPI.

The summit comes just three months after the release of Linux 2.4, which took more than two years to finish. The event drew hundreds of heavyweight programmers in the Linux kernel development community, notably Linus Torvalds, Alan Cox and Stephen Tweedie. The summit was funded by IBM, AMD and EMC.

Ts'o refused to specify a release date for Linux 2.5, but said it will be released within a nine to 18-month window--and faster than the Linux 2.4 kernel. "We did touch upon ways to speed up the development cycle," said Ts'o. "Linux 2.4 took a lot longer than we thought it might take. We're striving to shorten the development cycle."

While the kernel spec is still in the blueprint phase, programmers at the conference also discussed plans to support new HotPlug architectures to enhance the hot-swapping capabilities of Linux, and on the high end, to pursue enhancements for ccNUMA and SMP clustering, said Ts'o. IBM executives who attended the event, for instance, focused on Linux enhancements for ccNUMA architectures, he said.

Ts'o deflected a suggestion of impropriety between corporate funding of the event and participation in the kernel development by AMD and IBM. AMD merely paid for a Webcast of the event, Ts'o said, adding that other corporate funding paid for the event and for "travel scholarships" awarded to a handful of independent Linux developers. The event was co-sponsored by the Open Source Developer's Network and USENIX.

Analysts say the direction of the kernel is sound, but it's too early in the game to comment

"These goals are positive developments," says George Weiss, an analyst at Gartner. "We need to see how effectively Torvalds' group can incorporate these improvements, generate ISV enthusiasm for a standards base 2.4/2.5 kernel and see vendor-demonstrated real world benchmarks."

Another analyst notes that the challenge isn't getting corporate support, but managing the development in a timely manner.

"The issue isn't engendering more corporate support for Linux kernel development; in fact, every Unix vendor has a vested interest in supporting the development of the Linux kernel, and most already are active participants," says Stacey Quandt, associate analyst for Linux and Open Source software at Giga Information Group. "The challenge will be balancing a faster kernel development process with the addition of system software. The hard part is weighing best-of-breed technology from the Unix vendors against what is feasible to integrate into the Linux 2.5 development process."

At least one solution provider claims the kernel development has little impact on his business or his customers. The real focus, he says, is on forthcoming enhancements in the open-source arena for Web servers, file and print services and GUI efforts.

"The Linux 2.5 changes sound great, but most of the high-end things won't have direct impact on our customers. The kernel features really don't have much impact on end-users," says Tony Awtrey, vice president and director of integration at Anthony L. Awtrey Consulting, a solution provider for Unix, Linux and Windows NT in Melbourne, Fla. "Things like Apache 2.0 and Samba-TNG, as well as changes in Gnome and KDE, will more directly affect their experience using Linux."