View Single Post
Old Dec 23rd, 2005, 3:53 PM   #19
Mad_guy
Hobbyist Programmer
 
Mad_guy's Avatar
 
Join Date: Oct 2004
Location: Sandstorm, Techno Club
Posts: 239
Rep Power: 4 Mad_guy is on a distinguished road
Send a message via AIM to Mad_guy Send a message via MSN to Mad_guy
Quote:
Originally Posted by nnxion
Given the power of computers nowadays, microkernels can be a good choice. Although some would argue that the overhead would be coming from the lowest level, which wouldn't be nice.
Microkernels could possibly be the next method for creation of operating systems, yes. However, like I said, the driver problem isn't solved by microkernels, what if one driver is made to interfere with another important one? What if that driver doesn't need any interference but falls anyway? It is very possible people may start going to the microkernel solution in the future, but given the current stability of the quite impressive NT kernel Microsoft has made (actually, NT is more like a hybrid Microkernel, but it doesn't stray too far away from standard Monoliths, more Mono than Micro I'd say), I'd say Monolithic systems are just doing fine. DragonflyBSD is the first non-Mach BSD system to adopt the Hybrid kernel solution. But hybrids are far from perfect (hybrid kernels are simply kernels where some of the code is stored in the kernel (such as drivers) to increase speed, but things like the NT executive are user mode, even though it's a crucial part of the system.)

An interesting modus operandi to solve the Kernel Driver conundrum actually, is the thought of "No kernel" systems. Definately cool. Check out the TUNES project or Unununium. I can't really comment on Exokernels.
__________________
os: mac os 10.5.4
revision control: git
editor: emacs

site
Mad_guy is offline   Reply With Quote