Linux kernel 2.6 previewed

An explanatory summary that should be accessible for most people in the IT industry is available here, by Joseph Pranevich, and a rather more concise overview for the core minimalists and more technically inclined audience among us is available here, by Dave Jones.

Neither of these writers has gone out of their way to extol the virtues of the changes that will appear in the next version of the Linux kernel, but even a casual read makes it clear that big advances will show up.

Improvements in Linux 2.6 will benefit platforms ranging from embedded, near real-time devices to very large multi-CPU SMP and HPC systems, all using the same Linux kernel running on multiple system architectures.

Here are the headings in Joe Pranevich's excellent and detailed summary entitled "The Wonderful World of Linux 2.6":

The Story So Far...

Multiple Platform Support

Scaling Down -- Linux for Embedded Systems

Opteron Support - 64-bit Linux for the Consumer

Subarchitecture Support

Scaling Up -- NUMA and Bigger Iron

Linux Internals

Hyperthreading

Size Matters -- Scalability Improvements

Kernel Interactivity and Responsiveness

Module Subsystem - Device Drivers

Other Improvements

Unified Device Model

Kernel Object Abstraction

The System Filesystem

Core Hardware Support

Internal Device Busses

External Device Busses

Wireless Devices

Block Device Support

Storage Busses

Filesystems

Input / Output Devices

Human Interface Devices

Audio

Multimedia

Software Improvements

Networking

Network Filesystems

Miscellaneous Features

Security

Virtualizing Linux

Laptops

Legacy Support

Stuff At The Bottom

It's impossible to emphasize a single area of Linux advances, since all of these are likely to be important to someone. However, it is perhaps worth noting that several major functions of the Linux kernel have been completely rewritten. In fact, it's almost all revised. Quite a list!

You'll have to read the articles linked to see if something you'll care deeply about is being changed. But if you're using Linux and you haven't been involved in writing the 2.5 development kernel, now might be a good time to get involved with wringing out pre-production Linux 2.6 kernel code -- if only to find out whether it runs well on your platforms and works well with your applications, and possibly provide feedback to the kernel development team if you find any bugs, so they will get fixed.

The competition took about ten years to produce Windows Server 2003, but it's still insecure (as evidenced by the recent exploits discovered) and still pointy-clicky, to put it nicely. In contrast, Linux appears to be on track with massive revisions and significant advances to be delivered to all of us within about another six months or so, when it's ready. µ