Re: Is hibernation an argument for orthogonal persistance?

[Jonathan S. Shapiro]

Actually, hibernation may be an argument *against* orthogonal
persistence. The problem is that in the eyes of the user it is hard to
tell the difference, and therefore hard to see why orthogonal
persistence is a good thing to have.

The difference is that orthogonal persistence is more than just
persistence. It imposes an architectural discipline in which the kernel
state is just a cache. This is why you can have millions of processes
live simultaneously in EROS or Coyotos.

With hibernation, you haven't actually changed the system structure in
any fundamental way, so scalability remains limited by the kernel's
virtual address region.

shap


On Sun, 2006-02-12 at 15:29 -0800, Jeremy Shaw wrote:
> Hello,
> 
> Since laptops are now outselling desktops (atleast in the US)[1], I
> think it is safe to say that good hibernation support is a valuable
> feature of a modern OS.
> 
> It seems to me that hibernation (aka, suspending to disk) is really
> just a specialized usage of orthogonal persistence. If that is true,
> then it seems like it might be worthwhile to fully exploit
> persistencce, since many of the hard parts have to be done to support
> hibernation anyway.
> 
> One obvious difference is that orthogonal persistence needs to not
> interfere with normal operation of the machine whereas hibernation can
> take a few seconds to do its job.
> 
> Just a thought...
> 
> Jeremy Shaw
> 
> [1] http://arstechnica.com/news.ars/post/20050606-4971.html
> 
> 
> _______________________________________________
> L4-hurd mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/l4-hurd