____/ ed on Thursday 08 November 2007 23:50 : \____
> On Thu, 08 Nov 2007 19:47:16 +0000
> Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> wrote:
>
>> ____/ ed on Thursday 08 November 2007 14:41 : \____
>>
>> > On Wed, 07 Nov 2007 23:05:30 +0000
>> > Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> wrote:
>> >
>> >> Linux Boot Time Greatly Reduced, x86 Version of TP InstantBoot
>> >> Announced
>> >>
>> >> ,----[ Quote ]
>> >> | The boot loader extension transfers and expands the image to the
>> >> memory at | startup and returns the state of the InstantBoot driver
>> >> to what it was just | before power was shut off. The solution
>> >> reduces startup time through a | combination of these processes.
>> >> `----
>> >>
>> >> http://techon.nikkeibp.co.jp/english/NEWS_EN/20071107/141935/
>> >
>> > I think if one compiles the kernel so that it's monolithic without
>> > module loading things get going a bit quicker. The kitchen sink
>> > kernels that have a bunch of stuff that's not required will take
>> > longer to boot and compile.
>> >
>> > This might make it a bit easier to distribute for embedded systems
>> > though if the state is saved like this suggests.
>>
>> The company exaggerates when it comes to boot times (marketing talk)
>> because many distributions disabled some services to make booting
>> fast. OpenSUSE, for example, would take about 30 seconds to boot on a
>> fast PC.
>
> Yeah boot times don't mean much really, what's important is giving the
> user the ability to turn things on.
>
> What is important though, is that a desktop should include few daemons
> that start at boot. Instead gdm should spawn these if a user wants
> them (of course - the root user should enable this, or possibly
> configure this for the user). To give an example, if bob wants to work
> with apache, then the root user should set this. When bob logs in, gdm
> should spawn that for bob if it's not already running.
>
> I don't know if this is done already, if not give me a shout, maybe
> I'll knock it up.
>
> Boot times should only be important for a desktop. For servers, it's
> really not important, many of mine have been running > 1.5 years before
> hardware failures - it's no big deal.
The same goes for laptops. Hibernation makes it a non-issue and Linux doesn't
bleed and leak memory like a RAM-hungry and mis-managed Windows (requires
reboot to 'refresh').
> It is interesting though, offloading the state. I should read up more
> on this.
--
~~ Best of wishes
.oʍʇ sɐ buıɥʇ ɥɔns ou s,ǝɹǝɥʇ 'ɹǝpuǝq 'ʎɹɹoʍ ʇ,uop :ʎɹɟ
.oʍʇ ɐ ʍɐs ı ʇɥbnoɥʇ ı puɐ ...ǝɹǝɥʍʎɹǝʌǝ soɹǝz puɐ sǝuo .ɯɐǝɹp 1nɟʍɐ uɐ
ʇɐɥʍ 'ɥɥɥɐ :ɹǝpuǝq
|
|
