<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Re: Oracle Memory Utilisation in Operating System - HP-UX</title>
    <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741733#M944052</link>
    <description>Thanks for the help. &lt;BR /&gt;Although I am sending to you the memory utilisastion extract from:&lt;BR /&gt;UNIX95= ps -e -o ruser,vsz,pid,args | sort -rnk2 | grep bccs&lt;BR /&gt;&lt;BR /&gt;What you will see is that the system utilises 6.5 GB of memory. &lt;BR /&gt;&lt;BR /&gt;In the system there are 8 GB of Memory and 8 Processors.&lt;BR /&gt;&lt;BR /&gt;Is it OK to use SHMMAX 8GB i.e. full memory or i have to use 7GB&lt;BR /&gt;&lt;BR /&gt;Thanks a lot for the help&lt;BR /&gt;&lt;BR /&gt;George</description>
    <pubDate>Tue, 11 Jun 2002 08:28:19 GMT</pubDate>
    <dc:creator>George Nikoloudis_1</dc:creator>
    <dc:date>2002-06-11T08:28:19Z</dc:date>
    <item>
      <title>Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741730#M944049</link>
      <description>Dear Forum&lt;BR /&gt;&lt;BR /&gt;I have a problem with the Oracle memory utilisation. &lt;BR /&gt;&lt;BR /&gt;This problem is the following:&lt;BR /&gt;As you can see from the document glance_snapshot.txt and in applicaiton number (4) the memory utilisation  is 3.59GB physical and 190GB virtual. &lt;BR /&gt;&lt;BR /&gt;All the other applications have less than 1GB. &lt;BR /&gt;&lt;BR /&gt;In the init_db.txt file which is the init file of the same oracle database  the SGA is about 1.5GB. &lt;BR /&gt;&lt;BR /&gt;Each process of oracle reserve from 25MB up to 70MB and there are at least 100 such processes.&lt;BR /&gt;&lt;BR /&gt;I have also atached the kernel file in order for you to check if any parameter is over estimated or wrongly adjusted.&lt;BR /&gt;&lt;BR /&gt;Finnaly I am sending to you the swapinfo.txt file with the swap utilisation.&lt;BR /&gt;&lt;BR /&gt;My goal is if it possible to get some system indications about the memory utilisation and the parameters affecting this from oracle side. &lt;BR /&gt;&lt;BR /&gt;Also any changes or recommendations of the kernel in order the memory utilisation to be normalised are wellcome.&lt;BR /&gt;&lt;BR /&gt;I would like to thank you for reading my message.&lt;BR /&gt;&lt;BR /&gt;BR&lt;BR /&gt;George</description>
      <pubDate>Tue, 11 Jun 2002 06:34:40 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741730#M944049</guid>
      <dc:creator>George Nikoloudis_1</dc:creator>
      <dc:date>2002-06-11T06:34:40Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741731#M944050</link>
      <description>Hi,&lt;BR /&gt;&lt;BR /&gt;First and foremost, you wanted an SGA of 1.5 GB, but your shmmax is only 0X40000000 or 1073741824 (1 GB).&lt;BR /&gt;&lt;BR /&gt;You will need to increase this to at least 2 GB (OS also uses some shared memory).&lt;BR /&gt;&lt;BR /&gt;Hope this helps. Regards.&lt;BR /&gt;&lt;BR /&gt;Steven Sim Kok Leong</description>
      <pubDate>Tue, 11 Jun 2002 07:31:51 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741731#M944050</guid>
      <dc:creator>Steven Sim Kok Leong</dc:creator>
      <dc:date>2002-06-11T07:31:51Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741732#M944051</link>
      <description>Hi George&lt;BR /&gt;      It also looks like IO for the application in question is very high compared with the other applications. &lt;BR /&gt;  It may be that IO has become a bottleneck for this appl. Check the layout of the disk onto which this application processes do IO. Try to organise the disks in such a way that load is spread across different disks.&lt;BR /&gt;  Also as Steve has suggested increase the SHMMAX, otherwise you will be needing to shared memory segments for the application SGA which is not optimal.&lt;BR /&gt;  &lt;BR /&gt;...BPK...</description>
      <pubDate>Tue, 11 Jun 2002 07:43:04 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741732#M944051</guid>
      <dc:creator>Praveen Bezawada</dc:creator>
      <dc:date>2002-06-11T07:43:04Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741733#M944052</link>
      <description>Thanks for the help. &lt;BR /&gt;Although I am sending to you the memory utilisastion extract from:&lt;BR /&gt;UNIX95= ps -e -o ruser,vsz,pid,args | sort -rnk2 | grep bccs&lt;BR /&gt;&lt;BR /&gt;What you will see is that the system utilises 6.5 GB of memory. &lt;BR /&gt;&lt;BR /&gt;In the system there are 8 GB of Memory and 8 Processors.&lt;BR /&gt;&lt;BR /&gt;Is it OK to use SHMMAX 8GB i.e. full memory or i have to use 7GB&lt;BR /&gt;&lt;BR /&gt;Thanks a lot for the help&lt;BR /&gt;&lt;BR /&gt;George</description>
      <pubDate>Tue, 11 Jun 2002 08:28:19 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741733#M944052</guid>
      <dc:creator>George Nikoloudis_1</dc:creator>
      <dc:date>2002-06-11T08:28:19Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741734#M944053</link>
      <description>Hi George&lt;BR /&gt;   SHMMAX is the system-wide maximum allowable shared memory segment size.&lt;BR /&gt;    This means it is the maximum of a single shared memory segment. In your case as you need SGA of 1.5GB having SHMMAX value of 2GB would be sufficient.&lt;BR /&gt;  This parameter in assocation with SHMMNI , which is the system wide maximum number of shared memory segment control your shared memory allocations.&lt;BR /&gt;&lt;BR /&gt;...BPK...</description>
      <pubDate>Tue, 11 Jun 2002 10:16:33 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741734#M944053</guid>
      <dc:creator>Praveen Bezawada</dc:creator>
      <dc:date>2002-06-11T10:16:33Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741735#M944054</link>
      <description>I would suggest increasing the kernel parameters mentioned previously to 4g each:&lt;BR /&gt;&lt;BR /&gt;maxdsiz, maxdsiz_64bit, maxtsiz, maxtsiz_64bit, and shmmax.  &lt;BR /&gt;&lt;BR /&gt;Also, I would drop your log_buffer from 400m to 500k.  Setting it this high is probably the cause of alot of your problems.  Oracle doesn't recommend that this go over 1m.  Also, what kind of application are you running.  With 100 processes, a 10m sort_area_size will take 1g of memory alone.  &lt;BR /&gt;&lt;BR /&gt;The following are two notes in Metalink regarding sort_area_size.  If you do not have a Metalink account, I will copy them as attachments:  &lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=30918.1" target="_blank"&gt;http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=30918.1&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=109907.1" target="_blank"&gt;http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=109907.1&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;You should consider looking over all of the databases running on this machine for the same information.  &lt;BR /&gt;&lt;BR /&gt;Brian</description>
      <pubDate>Tue, 11 Jun 2002 23:48:52 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741735#M944054</guid>
      <dc:creator>Brian Crabtree</dc:creator>
      <dc:date>2002-06-11T23:48:52Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741736#M944055</link>
      <description>I would like to comment:&lt;BR /&gt;&lt;BR /&gt;1) Can you provide more detail about the log buffer of oracle in order to modify accordingly.&lt;BR /&gt;&lt;BR /&gt;2) Based on the parameter REDO_LOG_SPACE_REQUEST in the V$SYSSTAT, if this number is increased over an operational period the log_buffer should increased accordingly in order the value of the REDO_LOG_SPACE_REQUEST to get closer to zero&lt;BR /&gt;&lt;BR /&gt;3) Assuming that the Oracle processes are sharing the data and text from the shared memory segment of SGA then can you explain why there are proceses that have size from 25MB up to 96MB? Assume that the  SORT_AREA_SIZE is 10MB, please see the attached file.&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;</description>
      <pubDate>Wed, 12 Jun 2002 05:48:47 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741736#M944055</guid>
      <dc:creator>George Nikoloudis_1</dc:creator>
      <dc:date>2002-06-12T05:48:47Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741737#M944056</link>
      <description>1.  Personally, I look at "redo wastage" when looking for an oversized buffer.  I dont recommend much over 1m except in situations where there are a large number of dataloads happening, and even then it should be kept low.  &lt;BR /&gt;&lt;BR /&gt;2.  REDO_LOG_SPACE_REQUEST is an ok parameter to look at, as it should be 0 in most cases.  However, there are a number of other parameters that you can set other than this.  You can increase LOG_SIMULTANEOUS_COPIES to increase the number of redo latches available on the system.  You can also increase or decrease the size of the LOG_SMALL_ENTRY_SIZE depending on how large each transation that you are running is.  &lt;BR /&gt;&lt;BR /&gt;An article on Metalink that talks about working out redo latch and log buffer contention is below.  &lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1" target="_blank"&gt;http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;3.  My assumption in this case is that the processes in question are owning part of the shared memory.  The "SZ" parameter from 'ps' will not give the correct information in these circumstances, because it does not differentiate between private and shared memory areas.  &lt;BR /&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt;&lt;BR /&gt;Brian</description>
      <pubDate>Thu, 13 Jun 2002 04:31:03 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741737#M944056</guid>
      <dc:creator>Brian Crabtree</dc:creator>
      <dc:date>2002-06-13T04:31:03Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741738#M944057</link>
      <description>1.  Personally, I look at "redo wastage" when looking for an oversized buffer.  I dont recommend much over 1m except in situations where there are a large number of dataloads happening, and even then it should be kept low.  &lt;BR /&gt;&lt;BR /&gt;2.  REDO_LOG_SPACE_REQUEST is an ok parameter to look at, as it should be 0 in most cases.  However, there are a number of other parameters that you can set other than this.  You can increase LOG_SIMULTANEOUS_COPIES to increase the number of redo latches available on the system.  You can also increase or decrease the size of the LOG_SMALL_ENTRY_SIZE depending on how large each transation that you are running is.  &lt;BR /&gt;&lt;BR /&gt;An article on Metalink that talks about working out redo latch and log buffer contention is below.  &lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1" target="_blank"&gt;http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;3.  My assumption in this case is that the processes in question are owning part of the shared memory.  The "SZ" parameter from 'ps' will not give the correct information in these circumstances, because it does not differentiate between private and shared memory areas.  &lt;BR /&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt;&lt;BR /&gt;Brian</description>
      <pubDate>Thu, 13 Jun 2002 04:32:10 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741738#M944057</guid>
      <dc:creator>Brian Crabtree</dc:creator>
      <dc:date>2002-06-13T04:32:10Z</dc:date>
    </item>
    <item>
      <title>Re: Oracle Memory Utilisation</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741739#M944058</link>
      <description>1.  Personally, I look at "redo wastage" when looking for an oversized buffer.  I dont recommend much over 1m except in situations where there are a large number of dataloads happening, and even then it should be kept low.  &lt;BR /&gt;&lt;BR /&gt;2.  REDO_LOG_SPACE_REQUEST is an ok parameter to look at, as it should be 0 in most cases.  However, there are a number of other parameters that you can set other than this.  You can increase LOG_SIMULTANEOUS_COPIES to increase the number of redo latches available on the system.  You can also increase or decrease the size of the LOG_SMALL_ENTRY_SIZE depending on how large each transation that you are running is.  &lt;BR /&gt;&lt;BR /&gt;An article on Metalink that talks about working out redo latch and log buffer contention is below.  &lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1" target="_blank"&gt;http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&amp;amp;p_id=147471.1&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;3.  My assumption in this case is that the processes in question are owning part of the shared memory.  The "SZ" parameter from 'ps' will not give the correct information in these circumstances, because it does not differentiate between private and shared memory areas.  &lt;BR /&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt;&lt;BR /&gt;Brian</description>
      <pubDate>Thu, 13 Jun 2002 04:32:12 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/oracle-memory-utilisation/m-p/2741739#M944058</guid>
      <dc:creator>Brian Crabtree</dc:creator>
      <dc:date>2002-06-13T04:32:12Z</dc:date>
    </item>
  </channel>
</rss>

