cancel
Showing results for 
Search instead for 
Did you mean: 

in st02 some swaps are showing in red?

Former Member
0 Kudos

when i cheked in st02 swaps are showing red field description and intial records as below and we restarted sustem 10days back

Field definition     99,30    260.623    119.313      48,70   199.978    163.516      81,77 157.793    195.891

Initial records      32,39     13.249      1.442      20,60    49.994     30.125      60,26   70.118     89.991

which parameters i need to change? my platform is windows and oracle.and hit ratio is above 99.

Thanks in Advance.

Accepted Solutions (0)

Answers (3)

Answers (3)

Former Member
0 Kudos

Hi Arun,

The Hit ratio for Prorgam is 99% then it is well and fine. The swaps are more because of the memory pages are been swapped for some program that doesn`t use the frequently used data. Normally the memory size of each page is 8 KB ,so the swapping for the Program until 2000 is well and good.

Check for the program, batch job with the response time which consumes more time and swaps.

But for a single batch job or program if you start changing the parameter value, it would go beyond the limit. Also Check on program with variants having selection criteria with higher limits.

You can even evaluate with the EWA report.

Thanks & Regards,

Ahamed.

Former Member
0 Kudos

To find out which parameters  to change just dubbleclick on the parameter name in ST02.

E.g.  Initial records and choose Curren parameters.

Read the notes before changing. Somethimes there are dependensies to other parameters.

Also check ipc/shm_psize* sizes with E.g. sappfpar before restarting the instance(s).

Former Member
0 Kudos

It would be easier if we could see the column headers to see if you've ran out of space or entries. A screen shot is often easier on the eye.

The parameters are:

Field description buffer FTAB

rsdb/ntab/ftabsize        300000 kB   Size of field description buffer

rsdb/ntab/entrycount    200000      Max. number / 2 of table descriptions buffered

Initial record buffer IRBD

rsdb/ntab/irbdsize        19500  kB   Size of initial record buffer

rsdb/ntab/entrycount      200000      Max. number / 2 of initial records buffered

There should be no swapping in these buffers at all (for optimal performance).

Former Member
0 Kudos

Hi Tony

present values of parameters

The parameters are:

Field description buffer FTAB

rsdb/ntab/ftabsize        245000 kB   Size of field description buffer

rsdb/ntab/entrycount    199978      Max. number / 2 of table descriptions buffered

Initial record buffer IRBD

rsdb/ntab/irbdsize        7000  kB   Size of initial record buffer

rsdb/ntab/entrycount      199978      Max. number / 2 of initial records buffered

Thanks

Former Member
0 Kudos

Can you paste in a screen shot of the ST02 values? If not, SAP normally recommend increasing the buffer sizes by 10% until swapping is eradicated. However, with these values, I'd double the size of rsdb/ntab/irbdsize and increase rsdb/ntab/ftabsize to 300000.

You also need to check if you have sufficient entries.

Former Member
0 Kudos

Hi Tony,

please find scrrenshot of ST02

Thanks in Advance

Arun.


Former Member
0 Kudos

From the screen shot, we can observe:

NTAB Buffers

- Hit ratios above 99% should be fine.

- 'Field definition' is configured to max (300MB) in my system with no displacements. There's no harm doing the same in yours but the hit ratio is not too low (optimal is above 99.5% for all NTAB buffers).

- 'Initial records' is also configured to the max in my Windows system (60MB). I'd do the same in your system - it's only 60MB.

Program

- no issues with this buffer. Hit ratio is good (above 95%) and 10,000 per day is an acceptable number.

OTR

- Double it and increase the number of entries if they've been exausted. As the other poster stated. double click the buffer and then click on 'Current parameters', which gives you the relevant parameters.

Table Buffers

- These are also very performance relevant. High processing time for transaction steps can be an indicator for swapping causing problems. Use Table Call Statistics (transaction ST10) to check the maximum size calculated for the tables since startup. This can give you an idea of the buffer size but will grow as negative values are added.

Then as the other poster stated, execute sappfpar e.g. 'sappfpar pf=[DRIVE]:\usr\sap\<SID>\SYS\profile\[instance profile] check' and check for warnings/errors.

I recommend that you pick up a copy of SAP Performance Optimization Guide. It covers the buffers in detail and much much more.

Former Member
0 Kudos

Thanks

Former Member
0 Kudos

Did this resolve the issue? Would be interested to know if the swapping went away...