The increase in file size you see at the OS level is due to overflow.
With static files, what is "static" is the modulo -- it doesn't change
unless you resize the file.  The modulo determines the number of
"groups" in the file; each group is allocated a primary buffer of a set
size -- in your case, 16K.  When more data is written to a particular
group than can be held in the primary buffer of 16K the data is stored
in overflow.  Overflow buffers are NOT allocated to the file at creation
and must be added dynamically as needed.  At the OS level, overflow
buffers are appended to the file, extending its length.  This produces
the increase in size that you are seeing.

The FILE.STAT shows 62 groups in level one overflow.  The file has
largish records (9255 bytes on average) and a largish standard deviation
(28,809) so I'd expect a lot of overflow distributed unevenly.  Probably
some of the overflow in the 62 overflown groups is in fairly long
chains.  Probably some very large records were added to the file during
the two hour period.

HTH,

Jeff Fitzgerald
Fitzgerald & Long, Inc.
www.fitzlong.com

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Jeff Butera
Sent: Sunday, April 13, 2008 12:51 PM
To: [email protected]
Subject: [U2] Unidata file size

I'm running Unidata 7.1.8 on Solaris 9.  I have a static hashed file
with modulo 5009 and block size of 16K.  I didn't think static hashed
files would change size at the OS level, but this one is.  Am I just wet
behind the ears or do I just misunderstand static hashing?

Here's the file this afternoon:

-rwxrwx---   1 datatel  staff    102105088 Apr 13 12:32 WWW.STATE

and about two hours later:

-rwxrwx---   1 datatel  staff    121372672 Apr 13 14:44 WWW.STATE

Since I know someone might ask:

File name                             = WWW.STATE
Number of groups in file (modulo)     = 5009
Static hashing, hash type             = 0
Block size                            = 16384
File has 62 groups in level one overflow.
Number of records                     = 505
Total number of bytes                 = 4673742

Average number of records per group   = 0.1
Standard deviation from average       = 0.3
Average number of bytes per group     = 933.1
Standard deviation from average       = 9586.7

Average number of bytes in a record   = 9254.9
Average number of bytes in record ID  = 21.6
Standard deviation from average       = 28809.0
Minimum number of bytes in a record   = 51
Maximum number of bytes in a record   = 381892

Minimum number of fields in a record  = 1 Maximum number of fields in a
record  = 827
Average number of fields per record   = 80.9
Standard deviation from average       = 134.7


Jeff Butera, Ph.D.
Administrative Systems
Hampshire College
[EMAIL PROTECTED]
413-559-5556

"Working with them is like going to Las Vegas:
   sometimes you get lucky and sometimes you don't."
                      Dan Parker
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to