Kernel Improvements, Controversies Will Be Focus of LinuxCon Roundtable

By August 27, 2009Blog
Article Source Community-cation
August 27, 2009, 6:16 am

One of the more popular sessions at LinuxCon is sure to be the Linux Kernel Roundtable, featuring several notable kernel developers, including Jonathan Corbet and Linus Torvalds. Moderating the Roundtable is James Bottomley, Distinguished Engineer at Novell, Director of the Linux Foundation and Chair of its Technical Advisory Board. Bottomley is Linux Kernel maintainer of the SCSI subsystem, the Linux Voyager port and the 53c700 driver.

I recently interviewed Bottomley to find out what his goals for the Roundtable are, where he believes the Linux kernel is now, and where it’s going. You have a lot of responsibilities surrounding the Linux kernel, as a maintainer as well as your activities with the Linux Foundation. How do you prioritize your many efforts?

James Bottomley: Novell gives me 50% of their time to spend on Linux (both community and the Linux Foundation). Obviously there’s more work than will fit into that so it tends to spill over into my free time. I use a very interrupt-driven scheduling model: I prioritise what I have to do; stuff at the top (such as bug fix patches) tends to get the most attention; stuff at the bottom tends to get forgotten about. What drives you to be a Linux kernel developer?

Bottomley: I think I’ve always been fascinated by systems. I started out life wanting to do chemistry, then switched to physics because I enjoyed the idea of modelling the universe with mathematical systems better than the more recipe driven transformations of chemistry. While doing physics, I got into computational modelling, and from there to actually helping design computers to do the modelling… when I got out of university, Computing seemed to be the thing, and Operating Systems, as closed mathematical models of how a system should work also fascinated me. At that time in the UK, we didn’t have Academic BSD source licences, so Linux was the only system where you could actually see the operating system code, so I naturally gravitated towards it (I actually persuaded the Mathematical Physics department to use it as the basis for buying cheap PCs for computation before I left).

When I moved to NCR in 1997, they had an x86 based SMP computing platform (Voyager) that linux didn’t support, so I interested some of the hardware and OS engineeers in a stealth project to do a Linux port for the 2.4 kernel (strictly in everyone’s spare time). That got me well known in the community, and from there I was picked up by a Linux HA start up to make shared storage HA work, hence my current position as SCSI subsystem maintainer. When you moderate the kernel panel at LinuxCon, what will be the theme?

Bottomley: The themes will be current controversies and future enhancements, including touching on some of the process and other issues that are going to be brought up at the kernel summit, like are we degrading performance by insufficient consideration and testing of patches, what should we be doing to improve the quality and readability of the code (without churning the entire code base). What technologies or projects excite you when looking at the future of Linux?

Bottomley: Well, obviously at a maintainer level, the current trend to convergence of network and storage (with technologies like FCoE and SRP) is fun to follow. From a broader technology angle, the convergence of wireless and cellular networks is the most interesting, promising Internet everywhere and the convergence of netbooks, cellphones and PDAs. What are some of the challenges you believe Linux will need to address in the days ahead?

Bottomley: So there’s two aspects to this: the challenges the kernel itself faces and the challenge that Linux as a whole faces. I think for the former it’s pretty much sustaining our lead as the most agile versatile platform running efficiently on the widest variety of devices (from supercomputers to embedded hardware).

For Linux as a whole, I think the challenge is to bring open source innovation to the end user computing experience, particularly on the desktop. We have some brilliant desktop technologies, but a lot of them are based on existing Windows or Mac stuff. I think the next releases of GNOME and KDE are pushing us much farther along towards the leading edge here, and I’m interested to see what actually emerges as viable technology for the next generation of user interfaces.