Swap memory & CPU rules

Hi people,
I need the requirements for choice the amount of CPU and type, and the rule to calculate the amount of swap memory of productive system. The amount of physical memory is 12 GB RAM.
In the installation guide says that PHYSICAL MEMORY+ 14 GB= Swap memory. Is this avalaible for productive system.
More thanks.

Ruben this is quite a frequently asked quetion...
Basically if your system is Unix based the thumb rule is that SWAP should be 3 times your physical memory (been 12 gb it shouldn't be much of an issue). In windows 64bit usually 20Gb of swap (pagefile) is enough for most installations
There is info at help.sap.com about this.
Regards
Juan
BTW, why won't you trust the info on the installation guide?

Similar Messages

Swap memory

Hi,
Version :10.2.0.1 and 11.2.0.1
Os :Aix 5/6
We often getting swap issues.It reached 95%.we have 120GB real memory and 45 GB swap space.The db size is 2.5TB.Some time the db went to hang state and we restarted the db.
DB level
Is it possible to find A QUERY which memory using whether real memory or swap memory?How much amount from it?
OS level
Is it possible to find how much memory A PROCESS using from real memory and swap memory?
Thanks.

922884 wrote:
Hi,
Version :10.2.0.1 and 11.2.0.1
Os :Aix 5/6
We often getting swap issues.It reached 95%.we have 120GB real memory and 45 GB swap space.The db size is 2.5TB.Some time the db went to hang state and we restarted the db.
DB level
Is it possible to find A QUERY which memory using whether real memory or swap memory?How much amount from it?
OS level
Is it possible to find how much memory A PROCESS using from real memory and swap memory?
Thanks.SWAP is an OS operation & Oracle has NO knowledge about it.
OS moves "idle" processes to free up RAM.
do like below & wait for the 1 minute data collection to finish.
post FORMATTED results using tags
[code]
bcm@bcm-laptop:~$ vmstat 6 10
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
r b   swpd   free   buff cache   si   so    bi    bo   in   cs us sy id wa
0 0     24 41488 124584 2292224    0    0    97    39 398 865 8 3 86 2
0 0     24 37756 124600 2296736    0    0     0    21 794 1825 6 4 89 1
0 0     24 42804 124616 2296752    0    0     0    19 683 1510 3 2 94 1
0 0     24 43060 124624 2296756    0    0     0    18 738 1669 4 2 94 1
0 0     24 43316 124640 2296760    0    0     0    21 732 1641 4 2 93 1
0 0     24 44564 124656 2296768    0    0     0    16 642 1376 3 1 95 1
0 0     24 44564 124684 2296760    0    0     0    63 686 1524 2 2 95 1
0 0     24 39464 124828 2296780    0    0    22    31 695 1536 4 3 93 1
0 0     24 39464 124844 2296796    0    0     0    27 696 1577 4 2 93 1
0 0     24 39340 124860 2296796    0    0     1    26 781 1784 6 3 90 1
[/code]
(si+so) should be less than (bi+bo)
Handle:      922884
Status Level:      Newbie
Registered:      Mar 22, 2012
Total Posts:      44
Total Questions:      22 (16 unresolved)
Why so many unanswered questions?

MEMORY ,CPU USING TOP

Hi
I want to know about the memory usage,cpu usage and i want to check is there any io contention
i used top command and got the following output.
load averages: 6.31, 5.71, 5.45 03:00:10
206 processes: 199 sleeping, 7 on cpu
CPU states: 47.5% idle, 49.7% user, 2.8% kernel, 0.0% iowait, 0.0% swap
Memory: 128G real, 3218M free, 55G swap in use, 72G swap free
PID USERNAME LWP PRI NICE SIZE RES STATE TIME CPU COMMAND
21947 oracle 2 0 0 0K 0K cpu/52 6:34 8.29% oracle
8816 oracle 2 10 0 0K 0K cpu/49 629:43 8.15% oracle
8087 oracle 2 0 0 0K 0K cpu/50 578:07 8.13% oracle
16828 oracle 2 0 0 0K 0K cpu/12 721:16 8.11% oracle
9253 oracle 2 0 0 0K 0K cpu/55 677:58 8.11% oracle
5122 oracle 2 30 0 0K 0K cpu/51 161:09 3.77% oracle
13254 oracle 2 42 0 0K 0K sleep 5:11 1.50% oracle
11499 oracle 2 52 0 0K 0K sleep 3:36 1.25% oracle
12759 oracle 2 33 0 0K 0K sleep 3:37 1.24% oracle
19578 oracle 2 59 0 0K 0K sleep 0:08 0.70% oracle
23188 oracle 2 59 0 0K 0K sleep 0:03 0.55% oracle
23190 oracle 2 59 0 0K 0K sleep 0:29 0.50% oracle
24789 oracle 2 59 0 0K 0K sleep 0:02 0.49% oracle
22434 oracle 2 59 0 0K 0K sleep 0:02 0.48% oracle
21484 oracle 2 59 0 0K 0K sleep 0:04 0.45% oracle
can anyone please help me to interpret the above?
Thx

Hi ora,
the top output don't tell me diddly-squat except that your load looks high and it looks like your swap might be an issue (but it could have been just pasted badly).
What is your problem... Is your database performance slow? Then use something like an AWR report to track the issue.
Is your OS performance bad? Then preferably ask your sysadmin to track the issue, but you can use commands like iostat, sar etc. to track contention on the OS.
You can look at /proc/meminfo for more info on memory.
I will say a good starting point will be to give us more info, Oracle and OS version and state your problem clearly.
Regards

Memory / CPU questoin

Server top output
CPU states: 31.1% idle, 24.1% user, 26.1% kernel, 18.7% iowait, 0.0% swap
Memory: 16G phys mem, 2277M free mem, 2052M swap, 1688M free swap
Can i put 2 more DBs here, This server is already running 19DBs??

Sure.
You can put another 20 DBs there if you want. You just may not want all 20 additional instances working at the same time.
Every database instance requires memory and CPU. However, the AMOUNT of memory depends on how the DBA configured it, and the AMOUNT of CPU depends on the load put on by the users.
So if your additional 2 Databases have running instance requirements that are below what your server can handle, you will be OK.
If your additional instances need more than what is available on the server, you will get general slow down on all instances as everything goes into swapping and/or paging.
Then again, I'd be asking WHY you need more DATABASEs. In many cases, the same effect can be accomplished by adding more SCHEMAs and that has the advantage of not increasing the overhead required by a blank instance.
Monitoring, and making these judgement calls, is what being a professional DBA is all about.

M5000 Memory Config Rules

Hello,
Iv'e been reading the M5000 Service Manual, XSCF Manual and I'm unclear about the memory configuration rules. Our M5000 is configured as a single domain with half of the server populated. 2 CPU boards and 4 Memory Boards. Each board has 8 x 2GB dimms for 64GB total. We would like to add another 2 Memory Boards (each 8 x 2GB).
This is where the confiusion is, the Configuration Rules say each XSB must have 1, 2 or 4 Memory Boards. But in a UNI-XSB configuration is the whole server one XSB? Can I just add the new memory boards to the next two open slots on the other half? Or do we have to move one of the CPU Boards to the other half of the server? Or is this not possible at all?
Thanks for any info.
-Todd

Hi.
One XSB on M5000 must include:
- 1 or 2 CPU board ( 2 or 4 CPU corresponding)
- 1 - 4 MemBoard
- 1 IO board.
It's look like you server have only one IO board so you can confgiure only one XSB and can't add additional MemBoard.
M5000 can consist 1 or 2 XSB ( depending of configuration. Main limitation - number of IO boards)
You can buy 1 additional IO-board and 2 ( or requird number ) MemB.
Move one CPU board to another half of ther server and configure additional XSF.
XSB 0 ( uni-mode) : 1 CPU-Board ; 4 MemB ; 1 IO
XSB 1 ( uni-mode) : 1 CPU-Board ; 2 MemB ( new) ; 1 IO (new)
Regards

MBP retina swap memory/humming issue

So i have 2 issues(may or may not be related) going on with my Retina MPB.
1. Whenever i play any flash video(youtube for ex:), it just goes crazy with humming sound(fan I imagine), even though it doesn't get hot and temps are within acceptable range. I don't see the process(chrome flash plugin) taking more than 15 to 20% cpu at max...and memory is acceptable as well, but this sound just keep fluctuating and can't figure out why exactly.
2. Mayb this is related to 1, but why does a machine with 16GB ram use 2.2 of swap memory? or I am reading activity monitor wrong?
http://cl.ly/image/2O3O381l111v (Screenshot of activity monitor)
Any thoughts?
thanks

You can check if some or other corner inside is overheating with Marcel Bresink's free Temperature Monitor. Or splurge a bit on his Hardware Monitor so you can check on temps, fans and the works. http://www.bresink.de/products.html
The use of swap on a user-class computer with 16GB RAM is indeed odd. I've been using mine all week long, running VMs on Fusion, all of Office's apps, browsers, whathaveyou, and swap still remains at zero. Only time I saw some swap space allocated was when I ran Gimp, that insists on swapping explicitly to disk, even if it is just editing a piddly 80k JPEG. But even then, the swap area allocated was just a few tens of MB's, nothing like that huge 2GB area. Besides Flash, care to share what else is running in there?

How to Re Enable Swap Memory on Osx 10.9.2

Hi
Need help in to renable the Swap Memory on Osx 10.9.2.

You can test your memory system in terminal, open terminal (use spotlight to find it or go to utility folder in applications) and issue the following command, watch your activity monitor while it runs, when you've had enough, hit Control-C in the terminal window.
sudo memory_pressure -l warning
watch your pressure rise then hit
control-C

Memory & CPU Usages

Hi,
I'm a new user of SJSAS. Currently I�ve installed v9.1 on redhat. I would be helpful for if any body answers questions below-
1. Is it possible to set different limit for memory & CPU usages for different applications, if yes then how ?
2. How can I monitor memory & CPU usages by different application ?
Thanks

By "application", I'm assuming you mean different web/ejb/enterprise applications within a single SJSAS instance. If that's true, then the answer to your question is no -- all applications share the resources of the appserver.
On the other hand, if you need to, you can set up multiple instances of the appserver on the same machine, deploy one application to each instance, and control the resources that the appserver itself uses. How you limit the CPU use (and monitor) is then done at the OS level -- I don't know how it's done in Linux, actually. But in Solaris, you could use the Solaris Resource Monitor, or processor sets, or Zones/LDOMs, or any number of things to affect the resources available to the appserver; look for similar things in Linux.

10.1.3 memory/CPU usage problems

I recently migrated from JDev 9.0.5.2 to 10.1.3.0.4 and impressed with the new features and functionality.
However I've noticed the interface seems significantly slower (eg. code insight and simply inserting a new line of code is much slower), and the memory footprint seems to continually grow. Seeing it at >300M is not unusual.
There are also times when JDev seems to simply hang, and when I look a the JDev process in WinXP, the memory usage grows seemingly out of control as well as gobbling CPU cycles. My only option to kill the process. I'm using a 2.6G Pentium, 1G RAM, WinXP Prof and JDK 1.5.0_05.
Are there known problems with JDev's memory/CPU usage, or known options I can turn off to improve performance?
TIA

I don't have the memory monitor extension and have turned off all non-essential extensions but still see the memory-growth problem.
One odd thing I noticed is the JDev process on WinXP grows by 2-3MB each time a breakpoint is hit. I verified that same behavior doesn't exist in JDev 9.0.5.2 for the same project, and happens no matter which J2SE I select for the 10.1.3 project (JDK 1.4 or 1.5) or which runtime VM (hotspot or ojvm).

After closing Firefox to use IE8 or any other program, everything is very choppy and slow. I have to go into Taskbar, Processes, and close firefox.exe manually (which is using up all the memory/CPU). Then, and only then, does everything behave normally. W

After closing Firefox to use IE8 or any other program (like InterVideo to watch a DVD), everything is very choppy and slow. I have to go into Taskbar, Processes, and close firefox.exe manually (which is using up all the memory/CPU and should have closed before). Then, and only then, does everything behave normally. What is going on and what am I doing wrong or missing??
== This happened ==
Every time Firefox opened
== about a couple months ago and happens every time I close Firefox.

'''Kill Application'''
In Task Manager, does firefox.exe show in the '''Processes''' tab?
See: '''[http://kb.mozillazine.org/Kill_application Kill Application]'''
'''Causes and solutions for Firefox hanging at exit:'''
'''[[Firefox hangs]]'''
'''[http://kb.mozillazine.org/Firefox_hangs#Hang_at_exit Firefox hangs at exit (Mozillazine article)]'''
'''[[Firefox is already running but is not responding]]'''
'''Safe Mode'''
You may need to use '''[[Safe Mode]]''' (click on "Safe Mode" and read) to localize the problem. Firefox SafeMode is a diagnostic mode that disables extensions and some other features of Firefox. If you are using a theme, switch to the DEFAULT theme: Tools > Add-ons > Themes before starting Safe Mode. When entering Safe Mode, do not check any items on the entry window, just click "Continue in Safe Mode". Test to see if the problem you are experiencing is corrected.
See:
'''[[Troubleshooting extensions and themes]]'''
'''[[Troubleshooting plugins]]'''
'''[[Basic Troubleshooting]]'''
If the problem does not occur in Safe-mode then disable all of your extensions and then try to find which is causing it by enabling one at a time until the problem reappears. You have to close and restart Firefox after each change via File > Restart Firefox (on Mac: Firefox > Quit). You can use "Disable all add-ons" on the Safe mode start window.
'''''Other Issues''''': to correct security/stability issues
'''Update Java''': your ver. 1.6.0.19; current ver. 1.6.0.20 (important security update 04-15-2010)
''(Windows users: Do the manual update; very easy.)''
See: '''[http://support.mozilla.com/en-US/kb/Using+the+Java+plugin+with+Firefox#Updates Updating Java]'''
Do the update with Firefox closed.

ST22 Swap memory problem

Hi all,
I having problem in BI 7.0 server for swap memory goes to RED color... I know some variables and paramaters, but I am not sure how to calculate (depend on 12GB physical memory)
Screenshot here... if anyone know how to fix that I really appriciated
http://www.flickr.com/photos/38842895@N04/4538676180/sizes/o/
Thank in advance

Double click on the red highlighted swap, now click on current parameters.
Here you can see which parameters designates the memory for this buffer area. Once buffer is exceeded it begins to SWAP.
this is not a bad thing unless you are seeing performance problems. Most performance problems are far deeper than this.
You now need to research the overall memory usage and decide whether to increase the parameter and possibly decrease others.
use ST06 as well for overall usage amounts of swap.
There is much to understand but you can start by reading here:
http://help.sap.com/saphelp_nw70/helpdata/en/34/d9c8b9c23c11d188b40000e83539c3/content.htm
Edited by: Augustus Cristicini on Apr 20, 2010 10:05 PM

Memory/CPU glutton

Right after chancing to Firefox 33.0 CPU use jumped insanely compared to 32 or any earlier. Even when I'm not opening anything or there's no active loading CPU use doesn't go below 50% use, when opening pages it takes easily double compared to earlier versions. Pages load slower and rolling them down is un-even.
Basically this version needs some serious work to reduce the load it puts on the memory/CPU, no amount of new options/safety justifies this amount of waste.
KK

Went back to 32.3, CPU use went down to half, memory usage the same. Exactly what did you add to it to explain/justify that much memory/CPU need?
Tell me when you've fixed 33's massive glutting and I'll use it, this version isn't worth second look.

Help URGENLY needed- swap memory info

Hello Folks,
I am new to solaris. I am writing a process monitoring program,
where I need to display the swap memory details.
I am confused with the different figures I am getting for the swap reporting utilities such as:
-> swap
$>swap -s
total: 733544k bytes allocated + 62376k reserved = 795920k used, 489800k available
Also
$>swap -l
swapfile dev swaplo blocks free
/dev/dsk/c0t0d0s1 32,1 16 1050752 264512
/export/home/swap/file1 - 16 409584 290352
/export/home/swap/file2 - 16 409584 314320
-> df
$>df -k
Filesystem kbytes used avail capacity Mounted on
/dev/dsk/c0t0d0s0 1600563 1461964 90583 95% /
/proc 0 0 0 0% /proc
fd 0 0 0 0% /dev/fd
mnttab 0 0 0 0% /etc/mnttab
swap 487896 0 487896 0% /var/run
swap 732224 244328 487896 34% /tmp
/dev/dsk/c0t0d0s7 15269571 13722010 1394866 91% /export/home
-> sar
$>sar -r 1
SunOS 5.8 Generic sun4u 07/08/02
15:25:07 freemem freeswap
15:25:08 3184 969141
Which of the above can be used to compare my codes o/p?
I would appreciate your help in this regrads in form of some C sample code snippet(if possible).
Thanks,
Rajan

iPhones contain no user servicible parts. IF you were to open the device, you would void any warranty associated, and Apple would no longer offer ANY service on the device (including out of warranty service).
You may want to look at http://support.apple.com/kb/HT4083 regarding importing pictures and videos.

Netweaver performance swap memory

Hello,
I run Netweaver 7 BI on Windows 2003 Server X64. I have 8 Gb of Memory.
After few minutes, Windows swaps : jlaunch using 4 GB of swap memory and kernel 3gb of swap memory.
This causes to slow down when accessing to SAP.
Any ideas to optimize this ?
Regards,
Chea-Lie

Hello Heng,
Greetings!
The main problem is SAP uses its memory parameters to maintain the buffers and these buffers are maintained by the profile parameters.
To find which of these buffers are low you will have to ST02 transaction and find the ones which have much swaps. SAP memory is classified into many varieties like ABAP program buffer, Screen Buffer, Table Buffer, Calendar..etc.
Double click on each of the buffers and find the current parameters from the button provided on the corresponding page. Search for SAP notes that could give you details on profile parameter maintenance for increased performance. Also find out whether the SAP ABAP system shows a low performance only when some of the user defined programs are running.
Maintaining the correct profile parameter values providing suitable buffer space for the SAP system could optimize the server performance issues.
For the Java system, I think you are not deploying any packages through Jlaunch. Also check the Virtual Memory of the SAP server, you would require a minimum of 20GB virtual memory for the normal performance of a SAP server with ABAP and Java instances. You also have memory parameters to be maintained for the J2EE engine.
Award points if the suggestion was useful.
Thank you
Regards,
Vineeth

Increase swap memory

Hi guys,
I'm trying to run some strain calculations but my PC keeps running out of memory (first the RAM (4gb) and after that the swap memory (1GB), checked it with free). I know that for these particular calculations <8GB of memory should suffice. Does anybody know how to increase the swap memory from 1GB to 4GB without reinstalling arch linux?
Cheers,
Joris
Last edited by keizer (2010-04-10 23:00:14)

1) replace swap with swapfile
http://wiki.archlinux.org/index.php/Swap
2) set properly swappiness:
vm.swappiness = 100
number corresponds to % of swap than can be used
set
vm.vfs_cache_pressure=500
you can adjust these settings permanently via /etc/sysctl.conf or temporarily:
sysctl -w vm.swappiness="100"
these settings will not affect system performance if RAM memory is not exhausted but will help if swap is needed.

Swap memory & CPU rules

Similar Messages

Maybe you are looking for