1
Is the Windows NT Kernel becoming outdated?
Started by
Linoman
, Nov 19 2006 04:41 PM
8 replies to this topic
#1
Posted 19 November 2006 - 04:41 PM
Microsoft have come a long way with the NT kernal. From the days of Windows 3.1 and then Windows NT, Windows 2000, Windows XP, Windows 2003 and now finally Windows Vista.
I must say the NT kernal has improved over the years, proberly one of its best achievements is replacing the 9x kernel that home users used to use. Now how much more improvements can Microsoft make to the kernel? Will there come a time when it will reach its limits?
Do you think Microsoft will have to start with something new?
I must say the NT kernal has improved over the years, proberly one of its best achievements is replacing the 9x kernel that home users used to use. Now how much more improvements can Microsoft make to the kernel? Will there come a time when it will reach its limits?
Do you think Microsoft will have to start with something new?
#2
Posted 19 November 2006 - 09:07 PM
Well, Linoman, if you are asking an opinion, here's mine. All things reach a limit where they are no longer effective or useful. The NT kernal is no different. With the onset of Vista, I fear that MS has reached that expandable limit. The size of the OS alone begins to boggle the imagination. A horse cart, at one time, was a great medium for moving things. It still is as long as the load and cart are balanced. The NT kernal is the horse cart in this case. So much is now loaded onto the cart that the strain is beginning to show. The next version of Windows (isn't it Vienna?) should be/needs to be built on an entirely new basis. This may require some changes in hardware and software but that is not a surprise for new Windows version. As hardware becomes more powerful and especially SMP, Windows/MS will need to make the evolutionary changes to keep up. Perhaps current hardware/software and Vista can get along...for a while. But the advent of the new GPUs alone casts some doubt to that scenario. There was a time when hardware was the bottleneck in a system. I believe that it soon will be the OS...if that day has not already arrived...talker.
#3
Posted 20 November 2006 - 01:37 AM
Microsoft "Vienna" will probably not be built on the NT kernel. I think it'll be built on a kernel written in managed code, similar to Singularity.
#4
Posted 20 November 2006 - 06:12 PM
thanks for that, brew
#5
Posted 20 November 2006 - 06:18 PM
I love how in the link the viena date is between 2009 and 2012. LOL, what a date range! Knowing MS they will probably not release it until 2015!
#6
Posted 21 November 2006 - 04:24 AM
Thanks brwein, that was a good read. And 2009 to 2012, I don't think so. I believe the release date will have much correlation to adoption of Vista. If Vista is wildly successful then the release date can be pushed back. If, however, Vista bombs or say reaches only 60% of expectations then we can feasibly look for the release of Vienna much sooner...talker.
And the new OS will come in an easy open can...T.
And the new OS will come in an easy open can...T.
Attached Files
#7
Posted 21 November 2006 - 07:21 AM
Classic, talker.
I'll take easy open any day
I'll take easy open any day
#9
Posted 25 November 2006 - 12:48 PM
NT Kernel has been expanded to handle more and more tasks demanded by new OSs for years. I think that M$ should adopt a new kind of kernel architecture for Vienna. The fact is that any kernel is something to be as small as possible, real-time fast and highly confident; and i belive NT Kernel is reaching its limit regarding that.
I like the idea of microkernels, running only the most basic low-leval tasks and having a kernel-side extension to handle file systems, mini drivers, protocols stacks and so on... all inside a special protected memory area. Note that im not saying that such things should run on user-mode, but maybe a special-mode between kernel-mode and user-mode. This way even very low-level components could crash and be recovered by the kernel.
I like the idea of microkernels, running only the most basic low-leval tasks and having a kernel-side extension to handle file systems, mini drivers, protocols stacks and so on... all inside a special protected memory area. Note that im not saying that such things should run on user-mode, but maybe a special-mode between kernel-mode and user-mode. This way even very low-level components could crash and be recovered by the kernel.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users