Swaps in ST02
Hi all,
We just increased the program buffer from 850 MB to 950 MB 3 weeks back and then to 1050 MB a week back.The system looks normal for a few days and then again i can see the swpas increaing.We already have 16 GB of memory on a 64bit Win NT platform.The DB is db2 v 8.1.5......
So should i again increae it. also i read somewhere that increasing too much program buffer area willl lead to buffer fragmentation .
I am not sure . please suggest if it is ok to increase.
Also the Nametab ,Table definition, Short NTAB,Field definition, Initial records buffers are showing a lot of swaps ..
I looked at the relevant parameters:
rsdb/ntab/entrycount = 70000
rsdb/ntab/ftabsize = 62500 KB
rsdb/ntab/sntabsize = 2500 KB
rsdb/ntab/irbdsize = 30 000 KB
Which values should i change...please suggest.....
Hi Kalyan and Juan,
thanks.. for replying...
The program buffer is showing about 18000 swaps...
also the
Field definition buffer shows 151,000 buffer swaps,
Short NTAB = 19,418 buffer swaps and
initial records buffer = 54,158 buffer swaps.
We just increased the program buffer from 850 MB to 950 MB 3 weeks back and then to 1050 MB a week back.The system looks normal for a few days and then again i can see the swpas increaing.We already have 16 GB of memory on a 64bit Win NT platform.The DB is db2 v 8.1.5......
Should there be something that has to be changed in the DB2 side, or should i again increae it. also i read somewhere that increasing too much program buffer area willl lead to buffer fragmentation .
I am not sure . please suggest if and to what extent i can increase....
Also the Nametab ,Table definition, Short NTAB,Field definition, Initial records buffers are showing a lot of swaps ..
I looked at the relevant parameters:
rsdb/ntab/entrycount = 70000
rsdb/ntab/ftabsize = 62500 KB
rsdb/ntab/sntabsize = 2500 KB
rsdb/ntab/irbdsize = 30 000 KB
Similar Messages
-
Swapping in ST02 in Initial Record Buffer rsdb/ntab/irbdsize
Hello All,
I've been trying to do some tuning in ST02 and was hoping someone might be able to explain why I am getting such a low hit rate on the Initial Records Buffer, and the high level of swapping, when there appears to be plenty free space and plenty free entries.
Here are the stats I see.
Efficiency
HITRATIO % 42
HITS 9,645
REQUESTS 22,790
DB access quality % 42
DB access 13,136
DB access saved 9,645
Reorgs 0
Size
Allocated KB 8,750
Available KB 7,500
Used KB 1,053
Free KB 6,447
Directory entries
Available 10,000
Used 3,801
Free 6,199
Swaps
Objects swapped 9,335
Frames swapped 0
Resets Total 0
Any clues would be much appreciated.
Thanks
SteveI have the same issue and my parameter settings are currently:
rsdb/ntab/irbdsize - 41500
rsdb/ntab/entrycount 45000
Are there any other suggestions? See below for swap information.
Efficiency HITRATIO % 43
HITS 10,243
REQUESTS 23,731
DB access quality % 43
DB access 13,488
DB access saved 10,243
Reorgs 0
Size Allocated KB 42,906
Available KB 41,500
Used KB 4,717
Free KB 36,783
Directory entries Available 11,250
Used 9,759
Free 1,491
Swaps Objects swapped 3,728
Frames swapped 0
Resets Total 0 -
Hi Gurus,
I have been seeing lot of swaps occuring in ST02,in one our production servers.
I could see the swaps occuring on CUA buffers in huge volumes
Can any body give me any clue on how to go about it,interms of analysing the swaps and fixing the same
Your help would highly be appriciated
Thanks in advance
Rgds-
Venuhello
you will find some help here : [Online help about buffers|http://help.sap.com/erp2005_ehp_04/helpdata/EN/c4/3a6e98505211d189550000e829fbbd/frameset.htm]
"When a buffer has insufficient freespace or free directories, it has to swap objects out of the buffer in order to load a new object. The column Swap shows how many objects have been swapped out since system startup."
This means that you have to increase the size of the buffer to avoid swapping.
To find the relevant parameter (in RZ10) for a specific buffer => transaction ST02, double click on the buffer and then click on the button "Current parameters". Here is/are the parameter(s) to increase (be careful of some parameters ... your system can crash or not restart !)
After saving & activating the profile in RZ10, perform at os level the command (under <sid>adm) : sappfpar check to fix the memory problems.
Hope this will help you -
Dear Experts,
In Transaction ST02 there are heavy swaps in the following buffers:
Field description
Initial records
Program
CUA
Screen
Tables
Generic key
Single record
Export/import
How to reduce them; If parameters has to be changed Pl let me know the parameters to be changed and also sequence to be followed.
It would really be helpful if you could let me know how to calculate the parameter values, and sequence to be followed as ease.
Details:
OS: HP-UX 11.23
ECC 5.0
Oracle Release 9.2.0.6.0
Physical mem avail Kb 16,764,976
Regards
InjamuriHi,
Please click [buffers|http://help.sap.com/erp2005_ehp_04/helpdata/EN/c4/3a6e98505211d189550000e829fbbd/frameset.htm] for information.
For Table Buffer or TABL
zcsa/table_buffer_area u2013 for size of table buffer data area.
zcsa/db_max_buftab u2013 for directory entries u2013 one for every resident table
For Single key table Buffer or TABLEP.
rtbb/max_tables u2013 Directory Entries u2013 One for each table.
rtbb/buffer_length u2013 Size of data area
Program buffer
abap/buffersize u2013 Only parameter. No of directory entries are calculated automatically.
For Screen Buffer or PRES.
zcsa/bufdir_entries u2013 Directory size u2013 One per screen.
zcsa/presentation_buffer_area u2013 Total screen buffer size in KB.
CUA buffers
rsdb/cua/buffersize u2013 total Buffer in KB and no of directories are caluated by dividing the same with 2K.
Role and paging buffer.
rdisp/ROLL_SHM u2013 For role buffer
rdisp/PG_SHM u2013 For paging buffer
Calendar Buffer
zcsa/calendar_area
With Regards,
Krishna. -
Hi All,
Does anyone know what should I do if I found that my system has high buffer swaps. The hit ratio is good enough(99%).
Thanks & Regards,
CasperHi casper,
a high Hit ratio is not ALWAYS a good thing. Is it possible that a very large table is buffered constantly in the buffer, which is accessed frequently, as a result of which, you get a high hit ratio, but there is no space left for other tables.
Please check if the ratio of Buffer Gets : User Calls if it is greater than 15..
Ifi t is, then you have some expensive select statements running and you need to tune those.
Hope this helps.
Regards
Shantanu -
Hi all,
When i checked the swaps in st02 it is showing as red.What to do plz.Should i change the parameters.
Buffer Hitratio Allocated Free space Dir. size Free directory Swaps Database
[%] [kB] [kB] [%] Entries Entries [%] accesses
Nametab (NTAB)
Table definition 96.10 7,151 0 0.00 30,000 0 0.00 159,968 385,089
Field description 96.55 32,344 5,702 19.01 30,000 21,404 71.35 254,965 263,737
Short NTAB 99.74 3,438 2,137 85.48 7,500 5,803 77.37 0 1,697
Initial records 96.18 5,438 177 3.93 7,500 654 8.72 96,328 103,175
Program 93.68 500,000 1,322 0.28 125,000 111,378 89.10 161,705 626,010
CUA 99.94 8,000 4,836 70.38 4,000 3,475 86.88 0 530
Screen 99.86 25,391 18,314 72.89 4,500 4,016 89.24 0 516
Calendar 100.00 488 394 82.08 200 95 47.50 0 105
OTR 100.00 4,096 3,531 100.00 2,000 2,000 100.00 0 0
Tables
Generic key 99.99 99,023 9,072 9.44 13,000 998 7.68 12 59,612
Single record 97.10 100,000 75,927 76.12 1,000 876 87.60 0 171,530
Export/import 66.42 20,000 11,764 81.85 20,000 18,587 92.94 0 0
Exp./Imp. SHM 16.95 4,096 3,490 98.84 2,000 1,999 99.95 0 0Hi Ambarish,
These are the values which u can take as reference
rsdb/ntab/entrycount 30000
rsdb/ntab/irbdsize 4000
rsdb/ntab/sntabsize 2500
rsdb/cua/buffersize 5000
sap/bufdir_entries 10000
zcsa/table_buffer_area 50000000
zcsa/db_max_buftab 10000
rtbb/buffer_length 30000
rtbb/max_tables 500
Release 4.0/4.5:
rsdb/obj/buffersize 4096 (kB)
rsdb/obj/max_objects 5000 (buffer entries)
rsdb/obj/large_object_size 8192 (Byte)
As of Release 4.6:
rsdb/obj/buffersize 20000 (kB)
rsdb/obj/max_objects 20000
rsdb/obj/large_object_size 8192 (Byte)
extend the buffers to 30 -40% this may help you
Regards'
Zia -
Dear Friends,
I am facing problem with SWAPS in ST02.I know the soloution like restarting the sytem/increasing the parametrers.Can any body help me in this issue with any other solution.i already restarted the system And one more thing is there any limit for swaps for each buffer.i am giving my PRD server swaps here.
Nametab (NTAB) swaps
Table definitio 1,342,725
Field definitio 910,742
Short NTAB
Initial records 970,789
program 2,527,221
CUA 3,473,889
Screen 119,352
Calendar
OTR
Tables
Generic Key 4,696
Single record 1,085
Export/import 11,735,724
Exp./ Imp. SHM
Thanks&Regards,
chaitanya.Hi Chaitanya,
The only way to control the amount of swaps the system produce is to fine tune the system parameters, a system restart only clear the existing ones but does not prevent swaps for happening...
is there any limit for swaps for each buffer
Nope, it will swap as many times as it needs to...
Can any body help me in this issue with any other solution?
As mention above the only solution is to tune your parameters and keep your kernel updated, small amount of swaps in ok, but excesive swaping it can go in detriment of your system performance.
I can tell you that 2,527,221 swaps in your program buffer simply means that its probably too small and so on... you'll have to spend time looking into it.
This is a well documented issue... search, research, read, test...
Regards
Juan -
DB02 in ECC60 on V5R4 takes a lot of time to open!!!
Hi Guys,
We recently installed an ECC60 system(SR2) on V5R4M0. The installation went pretty well. We installed a double stack system with both ABAP+JAVA.
After the install, did the post install including creation of my new clients and SGEN. Scheduled all the house keeping jobs and they are running fine.
Now when we go to run DB02 to run it, it takes alomost like 80-90 seconds to run it. We thought, this will be the first time only, but that's not the case...Everytime I run it, it takes 80-85 seconds to run it.
All other transactions which we use, open in less than half a second.
This machine is 15 GB, 2.5 CPU, p570 machine. I have set up a dedicated shared pool of 7000 MB for this system out of that 15 GB and am running this system out of this 7GB memory pool. The Machine pool is also set as per that formula given in one of the SAP notes, 428855.
When I run DB02, there are almost no Non-DB faults most of the time, with an occasional fault of 150-200 sometimes in the shared pool, but that too goes away within a second...so its pretty much no fault at all the times. No swaps in ST02 as well or overflows.
I ran the ST05 trace but nothing of significance., except for at places in the trace it shows 45 seconds for executing/fetch from the table DB4INDEX_STAT.
What are we missing? Or is it like this only in ECC60?
We have other SAP systems on this LPAR as well and they run DB02, within like 5-6 seconds...
So what's so special with ECC60?? We are there on the latest PTF's...
Any suggestions...
Thanks
AbhiHi Abhi,
first of all, I would like to mention, that all you did seems to be very useful ))
That it takes pretty long, is normal and I have seen this at several customers. Now, you should see how often you use DB02, I didn't see any problem there. In oposite to 4.6c we have about trippeled the objects, so at least it has to process far more data. But, if it increased from 5 to 90s this is not the only explanation ...
I didn't analyze this SQL by now, but if you feel to need a solution you could do this yourself or open an OSS ticket on this behaviour ...
Regards
Volker Gueldenpfennig, consolut.gmbh
http://www.consolut.de - http://www.4soi.de - http://www.easymarketplace.de -
Memory issues while running a yearly report
Hi,
One of my consultant is trying to run an yearly report which consumes approximately 3 GB of size. But, the memory available at that point is only 2 GB on the server. The report generates a runtime error "SYSTEM_NO_TASK_STORAGE".
But, when I restart the server, all the buffers are cleared and there is 5 GB space after which the report runs to success. The number of program swaps in ST02 are more than 3000. The swap space assigned is 400 MB.
I believe that increasing the swap space can resolve this issue. But, I would like to know any other better solutions for this.
Thanks
VijayHi,
Of course, the best solution would be to add some physical RAM on your server.
Nowadays, I would say that using less than 8 GB RAM for SAP products leeds running into problems.
In my company, we are now going full 64bit with at least 16 GB RAM...
Bythe way 400 MB for the program buffer seems very low to me. On SAP R/3 4.7 we use 1500 MB. But this just slows down the system.
You should check your extended memory setting.
Regards,
Olivier -
Performance Tuning Best Practices/Recommendations
We recently went like on a ECC6.0 system. We have 3 application servers that are showing a lot of swaps in ST02.
Our buffers were initially set based off of SAP Go-Live Analysis checks. But it is becoming apparent that we will need to enlarge some of our buffers.
Are there any tips and tricks I should be aware of when tuning the buffers?
Does making them too big decrease performance?
I am just wanting to adjust the system to allow the best performance possible, so any recommendations or best practices would be appreciated.
Thanks.Hi,
Please increase the value of parameters in small increments. If you set the parameters too large, memory is wasted. This can result in paging if too much memory is taken from the operating system and allocated to SAP buffers.
For example, if abap/buffersize is 500000, change this to 600000 or 650000. Then analyze the performance and adjust parameters accordingly.
Please check out <a href="http://help.sap.com/saphelp_nw04/helpdata/en/c4/3a6f4e505211d189550000e829fbbd/content.htm">this link</a> and all embedded links. The documentation provided there is fairly elaborate. Moreover, the thread mentioned by Prince Jose is very good for a guideline as well.
Best regards -
Profile parameter related to DIA increment
Hi all,
If we increase # of DIA wp using RZ10, the init/max swap requirement is calculated automatically for us. What about the buffer size (listed on the left side of no.of work process area in RZ10, i.e.abap program (abap/buffersize)? What is the best practive, do we have to adjust those buffer or wait till swap value(ST02) becoming high?
Thanks alot.Hi,
The ABAP program buffer size is adjusted by default when you install the SAP, but it is not necessarily the ideal value for the parameter.
It stores the compiled executable versions of ABAP programs (loads). So if you observer the swaps i.e. if the number of swaps for the ABAP buffer between 24hrs is more than say 8000, the you need to increase the buffer size. Also if the hit ratio for the ABAP buffer is below 98% then you need to consider increasing the buffer size. Swapping is a normal process but the number of swaps in one day is the key so if that is high you will have to increase the buffer size.
Hope that helps, <removed_by_moderator>
Read the "Rules of Engagement"
Edited by: Juan Reyes on Nov 2, 2009 9:22 AM -
Hi Experts,
In our Development and Production systems , we have been Observed that ,
Lot of Swaps at Program Buffer (PXA) Especially Development systems ..
Some times we were experianced Lot of swaps Export and Import Buffers also
When we saw at ST02 There was less Space at program buffer and export and Import Buffers .. , and also extended meory use is bit high ..
Could you please let me know that . If system is running with more swaps ( in St02) , How can we avoid this Swaps ?
Is there any formula to fine tune buffer parameters like program buffer, table buffer , Export/Import ,Roll area,HeapMemory to make the system up and running with good peformance ?
and also Paging area .........?
Thanks & Regards,
Praveen.Hi,
Do you take any ofline backups of your developement system ??? if yes then is it weekly or monthly ????if it is weekly need not to worry as the buffer gets reset as system gets recycled.and if theres no offline backups then u need to monitor it atleast a month...then plan to tune the buffer.
Swaps is the number (counter) by which old objects in the buffer is over written by the new objects from database .
i would suggest to monitor the swaps for a certain period
Edited by: rajubhattacharjee on Jan 25, 2011 7:04 AM
Edited by: rajubhattacharjee on Jan 25, 2011 7:13 AM
Edited by: rajubhattacharjee on Jan 25, 2011 7:23 AM -
Memory buffer tuning - Why bother?
Hi
We are running ERP 6 on Win2003 64 bit, SQL2005 SP2 (2.6TB in size), 8 App servers with 16GB RAM (avg 60% used). As part of our housekeeping ,we are looking at addressing some of the swaps in ST02.
Currently we are experiencing swaps in the following buffers:
Nametab - Field definition: 150 000 swaps
Program: 300 000 swaps
Export/import: 500 000 swaps
Our current avg dialog response time is between 700-800ms per month (1400-1600 concurrent users).
Would increasing these buffers have any visable positive affect on the system or for the end-user? How important is it to address these buffer swaps? SAP does not report on it in the weekly/monthly EWA reports.
Regards,
CJHi,
You have quite a lot of swaps but you don't say if it is per day or since startup.
I just checked our own production ECC6 system and we have only about 600 to 1500 program buffer swaps on only 3 of the 6 app servers. zero swaps for other buffers.
But we get the same kind of monthly avg dialog response time (600 ms).
If you have time, you can tune your buffer sizes, it could only be better but you will not get a big improvement.
With a program buffer much too small, I 've already seen very bad performance but it is not your case.
Some times, I like to say : If it is not broken, don't fix it!
Regards,
Olivier -
How do I size application server RZ10 parameters
What are some of the guidelines for sizing application server. I have 3 app servers and a CI/DB. I have properly sized the CI/DB with parameters based on HW and expected usage.
Is there a guideline which parameters that I can reuse from the CI/DB instance?
I haven't found any app server RZ10 parameters that are only for app server.
Or can I simply copy parameters from CI/DB over to the app server and update the domain name for each app server?
Thanks
Weyland YutaniHi Weyland,
There are no specific application server parameters in SAP. You can reuse the parameters of your CI, but you probably have to increase the number of wp's and decrease the number of batch wp's on the CI.
When you are running a traditional CI, you should remove the rdisp/wp_no_enq from the app server profiles.
You should take the usage of your system into account when doing your sizing. Do you want to use one app server for batches or is the load spread equally over the servers. Do you accept users to work in the CI or not.
As a rule of the thumb, you can use the following statement for sizing:
On your CI/DB reserve the memory for the OS and the database, the rest can be allocated for SAP.
On your app server, reserve the memory for the OS (and maybe other applications) and give the rest to SAP.
Furthermore this will be an ongoing process. After a couple of weeks running with your new servers, you probably have to change some parameters. Check the amount of swaps in st02, load distribution, etc.
Kind regards,
Mark -
Hello all,
My Solution Manager system is very slow. I am running Solution Manager 7.01 (EHP1 SP04) on Solaris 10 with Oracle 11.2.0.1. It is running in a T2000 Solaris box with 32 GB of RAM. This box hosts two other PI 7.0 systems as well, that are not busy at all.
I do not have swapping in st02, all buffer hit ratios are over 99% and st04/st06 data look good as well
However, SMSY for example takes 4-5 seconds to run and many queries in workcenters (webdynpro for abap programs) take even more , 10-12 seconds
Is there any documentation on what to look for in a Solution Manager system? The usual tuning methods haven't worked for me
Also, how do you I tune the database? Do I put parameter values for OLTP or OLAP? I am asking because as part of the Solution Manager Diagnostics, I had to configure a BW client as well.
Many thanks
Andreas> My Solution Manager system is very slow. I am running Solution Manager 7.01 (EHP1 SP04) on Solaris 10 with Oracle 11.2.0.1. It is running in a T2000 Solaris box with 32 GB of RAM. This box hosts two other PI 7.0 systems as well, that are not busy at all.
>
The T-Server series is the worst choice for a SAP system. Those CPUs have lots of threads but this comes only into play when you have multithreaded applications (like Java). Since ABAP is single threaded and can't leverage the CPU power you will factually be even slower than a 5 year older classical SPARC v9 CPU.
This is a known "issue" and there's not really much you can do about it.
This was discussed before in the following threads:
Re: ECC6 - DEV & QAS Performance Very Slow
high CPU response times, altough low CPU utilization
Markus
Maybe you are looking for
-
hi sir, I am new to SAP, so plz guide me. in Paper industry this is the actual Business Process: In papers industry they have different varieties of paper production in which they are characterized by the GSM (thickness), sizes and grades. These
-
ORA-01017: invalid username/password; logon denied on Weblogic 9.0 cluster
I am experiencing the following exception in Weblogic 9.0 server cluster environment. The same code works fine in non-cluster environment. The problem happens every now and then and would go away temporarily after recycling the domains. The Oracle da
-
Hi Team, We have Windows 2008 AD infrastructure with Single domain Single Forest & 30 remote AD sites with RODC in it. We are planning for NTP server setup on Windows 2008 server in DMZ......can someone help me with steps for setup. What is best prac
-
Calling TaskFlows From Multi ViewController
hi,iam using jdeveloper 11.1.2.3.0 i have an application that contains: modelA and ViewContorollerA and modelB and ViewControllerB i am using Dynamic Tabs UI Shell Template with class Launcher i want to call the taskFlows in ViewContorollerA and Vie
-
I can´t open Photoshop CC through the Bridge
When clicking on a photo (right mouse click), Photoshop is not shown as option and thus can not be selected...?!?! Any idea or Suggestion is welcome. Support already suggested to install new user account and to open it from there - not succesful. Tha