[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: should readahead allow generic paths?



At 3:59 AM -0400 10/11/07, Jon Masters wrote:
>Yo,
>
>Seems to me that it'd be more useful if the readahead package
>configuration files (default/later) could use wildcards for path
>components such as firefox, python, etc. (so that you don't have quite
>the disconnect/lock-step requirement between these packages).
>
>Is there a reason why it's hard wired?

It's an attempt to not read files that weren't read normally.  Since
readahead already reads whole files when usually only parts were read
normally, it's already at a disadvantage.  Reading unused files would be
worse.

Readahead has the ability to make a new list of files just for your own
machine.

In any case, readahead does not improve boot time, and unless you can show
how to make readahead improve boot time, there is unlikely to be any work
done on making it keep reading all the files after updates.  It doesn't
make booting any slower if readahead can't find its files, or any faster if
it can.

Microsoft's approach, which is probably patented, involves tracking what
disk blocks are actually read and re-organizing the data on disk so that no
seeking is involved.  It speeds up booting.

Readahead's approach involves reading entire files, for any file that was
read from during boot.  Readahead sorts its list of files to minimize
seeking.  It doesn't speed up booting.

Without finding a way that readahead can speed up booting, work put into
optimizing readhead is just wasted.
-- 
____________________________________________________________________
TonyN.:'                       <mailto:tonynelson georgeanelson com>
      '                              <http://www.georgeanelson.com/>


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]