NEXTSTEP utilized the Mach kernel, one of the earliest examples of a microkernel and the standard by which other microkernels are measured. It was developed in part by Avie Tevanian, former head of software at NeXT and, later, Chief Software Technology Officer at Apple until March this year. Mac OS X, being an evolution of NEXTSTEP / OpenStep, also utilizes the Mach microkernel. Is that about to change?

Apple Matters suggests that Apple may be considering switching from Mach to a monolithic kernel, similar to that utilized by Linux and other UNIX implementations. While the microkernel is a robust model and offers advantages such as great stability and security, it is generally slower than the monolithic kernel - 10% to 50% slower, according to Apple Matters. As Wikipedia put it:

Regarding the above comment, it should be noted that since the release of Mac OS X v10.0 in March of 2001, Apple has been improving the performance of Mach through progressive releases of the operating system. It should also be noted that there are microkernel alternatives to Mach that offer higher performance. The L4 microkernel, for example, offers "massive real-world performance improvements" over Mach; inter-process communications over 10x faster than Mach have been demonstrated.

Last year Low End Mac suggested that Mach is OS X's Achilles' Heel and voiced hopes that amid the switch to the Intel processor, Jobs and company would see fit to transition the OS away from Mach, but pointed out the reality that such a move would seem nigh on impossible with Avie Tevanian at the software helm. Well, now he's gone. Is the move to a faster kernel part of the breath of fresh air that some suggest his departure will bring? Are we overestimating the benefits of leaving Mach behind, and underestimating the difficulty of such a transition?