Ted MacNEIL wrote:
It appears that VSAM can create a CA split without a CI split, see z/OS DFSMS
Using Data Sets Chapter 9.
See the RedBook De-Mystifying VSAM.
It tells us that splits are not to be feared.
I think you mean VSAM Demystified SG 24-6105. Good book, I read it
entirely few
Gene Hudders wrote:
[...]
I sure would hate to take too many CA splits in an on-line environment
especially if the CA split requires another extent for either the
data and/or index component.
That's why I wrote not so bad. g
Indeed CA split is painful (CI also, but the pain is CI/CA times
Ted MacNEIL wrote:
I'm trying to understand how it works. Analyze the guts of VSAM. I'm learning!
I've played with VSAM since before ICF (26 years).
I never intended to denigrate what you were doing.
I thought you were expousing a problem; I didn't realise it was a learning
experience.
Sorry
to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
Gene Hudders wrote:
[...]
I sure would hate to take too many CA splits in an on-line environment
especially if the CA split requires another extent for either the
data
Ted MacNEIL wrote:
think you mean VSAM Demystified SG 24-6105.
Yes. I read it when it first came out, and haven't looked at it since.
So, what I laughingly call a memory (old-timer's), couldn't retain the original
title.
But, the point is still: who needs to care about splits anymore?
5:30 AMTo: [EMAIL PROTECTED]: Re: KSDS CA split, no CI split
CICS Guy,I think it does, REPRO is batch and it uses sequential access (i.e.
SIS),and this is exactly the same access that is used during the initial loadof
a VSAM cluster, and it is at SIS time that VSAM applies the free
spacevalues
I sure would hate to take too many CA splits in an on-line environment
especially if the CA split requires another extent for either the data and/or
index component.
That's the whole point of Demystified.
The only cost of the split is at 'split time'.
And, with today's disk and CPU capacity,
think you mean VSAM Demystified SG 24-6105.
Yes. I read it when it first came out, and haven't looked at it since.
So, what I laughingly call a memory (old-timer's), couldn't retain the original
title.
But, the point is still: who needs to care about splits anymore?
Regardless of the reasons
I'm trying to understand how it works. Analyze the guts of VSAM. I'm learning!
I've played with VSAM since before ICF (26 years).
I never intended to denigrate what you were doing.
I thought you were expousing a problem; I didn't realise it was a learning
experience.
Sorry about that.
-
Too
In a message dated 12/5/2007 12:07:04 P.M. SA Western Standard Time,
[EMAIL PROTECTED] writes:
Ted MacNEIL wrote:
It appears that VSAM can create a CA split without a CI split, see z/OS
DFSMS Using Data Sets Chapter 9.
See the RedBook De-Mystifying VSAM.
It tells us that splits are
Subject
Re: KSDS CA split, no CI split
I don't think so, from what I recall, free space is only reserved during
an initial load (maybe mass inserts too?)
From: IBM Mainframe Discussion List on behalf of Michael PoilSent: Mon
12/3/2007 12:05 PMTo: [EMAIL PROTECTED]: Re: KSDS CA split
It appears that VSAM can create a CA split without a CI split, see z/OS DFSMS
Using Data Sets Chapter 9.
See the RedBook De-Mystifying VSAM.
It tells us that splits are not to be feared.
-
Too busy driving to stop for gas!
--
. [EMAIL PROTECTED]
Sent by: IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
05/12/2007 18:00
Please respond to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
Gene Hudders wrote:
[...]
I sure would hate to take
/
--
R.S. [EMAIL PROTECTED]
Sent by: IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
04/12/2007 11:08
Please respond to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
Michael Poil wrote
Radoslaw,
Have you tried loading the data in reverse key order? (10, 9, 8, 7, etc)
Chuck
-Original Message-
From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of R.S.
Sent: Monday, December 03, 2007 11:03 AM
To: IBM-MAIN@BAMA.UA.EDU
Subject: Re: KSDS CA split, no CI
Michael Poil wrote:
RS,
Just remembered a VSAM gottcha. After initial load yit is better to
ALTER the dataset to use FSPC(0 0) if you are using Sequential Insert
Strategy so as to reduce the potential for CA splits.
Mike,
Bingo!
I did the following:
DEF CL ...FSPC(15 15)
REPRO (2
David Betten wrote:
Try increasing your index CI size. Sometimes when the keys are very unique
you get a situation where the index CI can't hold enough pointers for an
entire data CA. That might be causing the CA splits.
I was thinking about it, but the key doesn't seem to be poorly
Please respond to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
KSDS CA split, no CI split
The following scenario:
KSDS defined with RECSZ(100 100), FSPC(15 15), CISZ(4096). Records
loaded, with evenly distributed key values (i.e. start 1, increment
Michael Poil wrote:
RS,
Can you provide the LISTCAT output please.
Michael, please find output below.
I played with Index CISZ 1.5 - 18kB. I always get 3 CA splits.
I also played with CA freespace: 0-15 % gives 3 CA splits, 30-50% - 4 CA
splits, 60% - 5 CA splits.
DEF CL
.
Sent: Sunday, December 02, 2007 6:12 PM
To: IBM-MAIN@BAMA.UA.EDU
Subject: KSDS CA split, no CI split
The following scenario:
KSDS defined with RECSZ(100 100), FSPC(15 15), CISZ(4096). Records
loaded, with evenly distributed key values (i.e. start 1, increment 10).
2 records.
Then I inserted 500
Larry Crilley wrote:
I've seen when an INDEX CI gets full (can not put any more compressed keys
into the INDEX), a CA SPLIT can occur. Since the sequence set can not hold
the new key, the CA must be split.
Try loading your dataset and run an IDCAMS EXAMINE. EXAMINE has been
updated in the
I can think of 2 posible causes,
you do not have sufficient space in the INDEX component for any new CI's
therefore
1. any new records, have to go to a new CA ie CA split
2. the more CA's you have (ie increase free space for the CA which means the
data is loaded across more CA's) the more CA's
/
--
R.S. [EMAIL PROTECTED]
Sent by: IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
03/12/2007 15:16
Please respond to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
Larry Crilley wrote:
I've
Mike Bell wrote:
I can think of 2 posible causes,
you do not have sufficient space in the INDEX component for any new CI's
therefore
1. any new records, have to go to a new CA ie CA split
2. the more CA's you have (ie increase free space for the CA which means the
data is loaded across more
/
--
R.S. [EMAIL PROTECTED]
Sent by: IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
03/12/2007 15:16
Please respond to
IBM Mainframe Discussion List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
List IBM-MAIN@BAMA.UA.EDU
To
IBM-MAIN@BAMA.UA.EDU
cc
Subject
Re: KSDS CA split, no CI split
Mike Bell wrote:
I can think of 2 posible causes,
you do not have sufficient space in the INDEX component for any new
CI's
therefore
1. any new records, have to go to a new CA ie CA split
2
Radoslaw,
Have you tried loading the data in reverse key order? (10, 9, 8, 7, etc)
Chuck
-Original Message-
From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of R.S.
Sent: Monday, December 03, 2007 11:03 AM
To: IBM-MAIN@BAMA.UA.EDU
Subject: Re: KSDS CA split, no CI
I don't think so, from what I recall, free space is only reserved during an
initial load (maybe mass inserts too?)
From: IBM Mainframe Discussion List on behalf of Michael PoilSent: Mon
12/3/2007 12:05 PMTo: [EMAIL PROTECTED]: Re: KSDS CA split, no CI split
RS,Just remembered a VSAM
The following scenario:
KSDS defined with RECSZ(100 100), FSPC(15 15), CISZ(4096). Records
loaded, with evenly distributed key values (i.e. start 1, increment 10).
2 records.
Then I inserted 500 evenly distributed records (start=5, increment=100).
Listcat shows no CI splits and 3 CA
Try increasing your index CI size. Sometimes when the keys are very unique
you get a situation where the index CI can't hold enough pointers for an
entire data CA. That might be causing the CA splits.
Have a nice day,
Dave Betten
DFSORT Development, Performance Lead
IBM Corporation
email:
30 matches
Mail list logo
