Re: Index read performance

Charles,

File was changed to a Table and LFs were changed to indexes when file was
partitioned.

Both of them have the same parms as far as I can see:
MAINT *IMMED
UNIT *SSD
ACCPTHSIZ *MAX1TB
KEEPINMEM *NO
PAGESIZE *KEYLEN
25 (for the second file this is 33)
707


On Thu, Jan 28, 2016 at 10:24 AM, Charles Wilt <charles.wilt@xxxxxxxxx>
wrote:

Ok so 2μs for SETLL on the original "index" vs. 87us or 197us on the
replacement "index".

Are these actual SQL indexes? Or are they DDS logical files?

Regardless, any differences shown between them by DSPFD other than the

key?

Specifically:
UNIT *ANY
ACCPTHSIZ *MAX4GB
49
541
KEEPINMEM *NO

Charles

On Wed, Jan 27, 2016 at 6:52 PM, Vinay Gavankar <vinaygav@xxxxxxxxx>
wrote:

The data was gathered using performance tool (i-doctor I think). With

the

old file the elapsed time for SETLL operation (there is only one in the
program) took an average of 2 microsecs. With new file once it took 87

ms

and another time 197 ms.

The 3 tests were run on different days, I can't really comment on the
system load at each time. But between the second and third tests there

was

no code change, so it SEEMS that this is related to other factors, but

I

am

not sure.

On Wed, Jan 27, 2016 at 12:40 PM, Charles Wilt <charles.wilt@xxxxxxxxx

wrote:

Define substantial...are you looking at total wall-clock run time,

CPU

usage or is the program instrumented so that you can see how long

each

loop

takes?

I wouldn't expect a difference. How many time did you run it? Is it
possible that the first index was in memory? Use SETOBJACC

POOL(*PURGE)

to

ensure that neither index is in memory between runs.

Are you sure workload on the system was comparable between runs?

Charles

On Wed, Jan 27, 2016 at 11:01 AM, Vinay Gavankar <vinaygav@xxxxxxxxx

wrote:

Hi,

We have 2 indexes on a Table. First index is Key1, Key2 (Desc).

Second

index is Key1, Key2 (desc), key3 (desc), key4 (desc).


Key1 is 18 char, key2 & key3 are 8 numeric and key4 is 6 numeric.

We had a program which was using the first index to read all

records

for

one value of Key 1 (Setll and Reade on Key1). That program was

changed

to

use the second index (as they are planning to get rid of the first

index),

and a substantial performance degradation was observed.

Is this behavior to be expected due to the additional keys of the

second

index?

The Table is huge and partitioned. There is no Implicit Access path

sharing

in either files.

Any suggestions to improve performance would be appreciated.
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L)

mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L)

mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.