HP3000-L Archives

June 1998, Week 4

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Wirt Atmar <[log in to unmask]>
Reply To:
Date:
Tue, 23 Jun 1998 17:04:51 EDT
Content-Type:
text/plain
Parts/Attachments:
text/plain (59 lines)
Bill writes:

> Wirt wrote:

> >2. Doing this will keep your fragmentation problems to a minimum.
>  >
>
>  I have to disagree here.  While this type of balancing is a Good Thing, it
>  won't actually have much effect on fragmentation.  Fragmentation usually
>  occurs when files are built, purged, built, purged etc. with "holes" being
>  created throughout the disk.  This will still happen regardless of where
>  the initial files are allocated.  What will happen with Wirt's suggestion
>  is that disk allocation will be evenly balanced between drives of unlike
>  sizes.  Again, this is a Good Thing.

Bill may have misread my intention -- perhaps due to the language I chose. I
didn't mean to imply that disc balancing would elimination fragmentation. All
I meant to say was that properly balanced discs would minimize fragmentation
problems.

I have always thought of disc fragmentation as a simple thermodynamic analogy
to boiling water. The basic system files that you LOAD onto your machine tend
to be extremely stable and stay where they were loaded for the entire time
that you run your machine on this particular LOAD/RELOAD. These files are like
liquid water. And cold water, at that.

Further, for the first few minutes right after a LOAD, all of the files,
regardless of their permanency, are maximally compacted onto the disc and your
free space is just one large continuous block. This is much like the calm
surface of cold water -- and the free space is much like the air above the
water.

But, as soon as the machine begins to be used, as Bill says, files will begin
to be built and purged constantly, much like water beginning to boil. Now, at
any given time, there will be segments of active files pushed out into what
was previously pristine free space.

If you have a great deal of free space (the air-like quality in this analogy),
the boiling and bubbling of files coming into existence and disappearing is
constrained to pretty much that region right above the surface of the
permanent system files. The largest block of free space will remain pretty
much untouched through time.

But if you don't have enough free space (the cool water level is pretty much
up to the top on any one disc before you begin), the boiling nature of file
creation and extinction is going to leave what available area exists
honeycombed with current file segments -- and ultimately with free space
sections of the disc too small to be of use for the file that needs to be
created at the current instant.

But, if you do have an excess of space, and all of the discs are
proportionally balanced to their physical sizes, then that "boiling" region
where files are being created and purged will also be proportionally
distributed among the discs, leaving all of your discs with maximized free
spaces and thus the greatest liklihood of being able to build new file
segments when called upon to do so.

Wirt Atmar

ATOM RSS1 RSS2