HP3000-L Archives

April 1999, Week 2

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:
Denys Beauchemin <[log in to unmask]>
Reply To:
Date:
Mon, 12 Apr 1999 09:45:10 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (71 lines)
X-no-Archive:yes
It is my contention that you are doing counter-productive things.  There 
are only 3 acceptable block sizes in my opinion:  512, 1024 and 2048.  You 
will notice that 512 and 1024 are factors of 2048.  MPE/iX is page-based 
and a page is 4096 bytes (2048 16-bit words).  When the disk is read, pages 
are brought into memory.  By using 512, 1024 or 2048, you are assured that 
complete Image blocks are brought in every time.

I would try to minimize wasted space by looking at which of the 3 sizes 
reduces wasted space.  I would stay away from odd, but valid sizes, and I 
would never grow a block size bigger than 2048, unless there was absolutely 
no other way.

Kind regards,

Denys. . .

Denys Beauchemin
HICOMP America, Inc.
(800) 323-8863  (281) 288-7438         Fax: (281) 355-6879
denys at hicomp.com                             www.hicomp.com



-----Original Message-----
From:   [log in to unmask] [SMTP:[log in to unmask]]
Sent:   Monday, 12 April, 1999 9:25 AM
To:     [log in to unmask]
Subject:        DB blocking factor education.

I've been attempting to discover the "best" blocking factors to use on by 
datasets.  My goal is to reduce the wasted space in each block.  I've setup 
an Excel spreadsheet which processes each block size, and calculates the 
percent waste.  I'm using the "Solver" feature to determine the valid block 
sizes, and to show me the one with the minimum percent waste.
Could someone please let me know if these are the correct calculations? 
 They don't match what DBGeneral indicates as the block size to use for 
minimum disk.  In the spreadsheet, the "Rounded Block Size" is the "true" 
block size.
Bit Map Words Required
ÎILING(Blocking_Factor/16,1)
Block Size (Buffer Length)
=Bit Map Words Required+(Blocking_Factor*Media Length)
Number of 128 Word Blocks
ÎILING(Block Size/128,1)
Rounded Block Size
=Number of 128 Word Blocks*128
Wasted Space (Words) per Block
=Rounded Block Size-Block Size
Percent Waste
=Wasted Space/Rounded Block Size*100
I'm imposing the following restrictions:
Blocking Factor must be an integer.
Blocking Factor must be greater than or equal to 1.
Rounded Block Size must be less than or equal to 2560.
Rounded Block Size must be greater than or equal to 128.

For instance, on a dataset with a media length of 262, DBGeneral is 
indicating that the block size for minimum disk is 896, with a 12.2 percent 
waste.  My spreadsheet shows that if I use a "Rounded" block size of 896, 
there is 12.17 percent waste.  My spreadsheet is showing that the block 
size for minimum disk is 2432, with a 3 percent waste.
Does anyone know if the above calculations are correct, or why DBGeneral is 
reporting different numbers?
Thanks in advance,
David N. Lukenbill

--------------------------------------------
[log in to unmask]
Raytheon Missile Systems Company, Louisville

ATOM RSS1 RSS2