Subject: | |
From: | |
Reply To: | |
Date: | Wed, 17 Sep 1997 18:33:59 -0400 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
The file must have been in bad shape to think that re-indexing would buy
you
1000 percent improvements
Has anyone tested the disc extent/blocking usage in flat file equates ?
As with KSAM, they are no longer needed ....so they say....
>----------
>From: Exequiel R. Sevilla III[SMTP:[log in to unmask]]
>Sent: Wednesday, September 17, 1997 8:59AM
>To: [log in to unmask]
>Subject: [HP3000-L] Speed and KSAM
>
>Greetings.
>
>Yesterday I was modifying some data in a large-ish KSAM file when I decided
>to
>see what happened if I changed the blocking factor from 1 to 5. I expected
>not
>much difference, since the box is a 969-220 with MPEiX 5.0 RELEASE: C.50.03
>(an
>aside - I too find the Release/blah blah blah codes confusing)....I certainly
>didn't expect to save any disc space...and I believed the I/Os were supposed
>to
>not be affected. Much to my surprise, performance increased substantially -
>a
>serial read with Quiz improved from 1,052 CPU seconds to 97 seconds for 1.2+
>million records.
>
>I dropped a note to quote Paul Christidis, in case he would be able to
>provide
>some advice/explanation (not that I mind the improvement). He noted:
>
>>I understood that the 'blocking factor' parameter is *ignored* under MPE/iX,
>>it
>>is 'carried along' only for compatibility reasons in the event that one may
>>want to 'transport' data to an MPE/V machine. I just reread a section from
>>Eugene's book about MPE/XL disc files and he concurs.
>
>I've been able to replicate the results using a smaller KSAM file (82,000
>records - dropped from 69 seconds to 7) and on another large-ish one:
>
>Before:
>
>FILENAME CODE ------------LOGICAL RECORD----------- ----SPACE----
>SIZE TYP EOF LIMIT R/B SECTORS #X MX
>
>HR332KD KSAM 50B FA 999781 1300000 1 196960 * 32
>HR332KDK KSAMK 128W FB 247426 247426 1 247440 22 32
>
>After:
>
>FILENAME CODE ------------LOGICAL RECORD----------- ----SPACE----
> SIZE TYP EOF LIMIT R/B SECTORS #X MX
>
>HR332KD KSAM 50B FA 999326 1300000 5 196608 99 8
>HR332KDK KSAMK 128W FB 144450 144450 1 144464 22 8
>
>Using Quiz to serially read the file and produce a frequency cost 816 CPU
>seconds on the before, only 80 on the after.
>
>Any ideas why I've suddenly received such a speed increase?
>
>Kelly
>
>Exequiel R. Sevilla III (aka Kelly)
>Hughes Technical Services Company
>8614 Westwood Center Drive, Suite 950
>Vienna, VA 22182
>----------------------------------------
>703/918-6916 (voice) 703/918-6945 (fax)
>----------------------------------------
>e-mail: [log in to unmask]
>----------------------------------------
>Wind is like the air, only pushier.
>----------------------------------------
>
|
|
|