Huge Pages
In x86 Architecture, memory pages are like textbook pages where CPU use those pages for reading and writing them in virtual memory for addressing computing requirement. Technically, I can say it is a fixed size block of virtual memory which is defined as pages that has corresponding entry in page table and further this page table map the pages with fixed size memory frames of physical memory or also known as RAM. Operating Systems like Linux are capable of managing the physical memory with the help of virtual memory management at OS level.
In Linux default memory page size is 4KB. Although 4KB page size of memory is sufficient for running all types of Application but problem arises when you use Oracle database as Oracle uses its own method for memory management called SGA or Shared Global Area which supports large memory pool. Reason behind for SGA to support large memory pool is because it supports multiple components of database like database buffer, cache, shared pool, large pool etc. therefore using SGA with default memory page size would not be advisable as it will toll on Linux system in terms of high processing and wait time for accessing the memory.
Thus, to optimize performance and efficiency, it is often recommended to configure the large memory page appropriately when using Oracle database and the SGA. In Linux Huge page support larger memory pages sizes such as 2MB or 4MB, it enables the Linux to handle the large memory requirements of the SGA and reduce the overhead associated with accessing the memory. Huge pages cannot be swapped out to disk thus it also help in improving the performance.
Steps to configure the Huge Pages
In this blog I will explain the steps on how to configure the Huge Pages in Linux system. Note these steps require Oracle database downtime therefore perform this activity with an adequate database downtime window:
- Run the below command to check if Linux kernel supports Huge Page settings:
- Increase the memory lock limit to 90% of RAM (in KB value) in /etc/security/limits.conf file as this is required for supporting Huge Page size. As I have shown in below example my Linux system has 2.6 GB of memory and I have increased the memory locking value to 2453667 KB which is 90% of 2.6GB memory in KB size:
- Now login with oracle user id and verify the new memory locking setting:
- Now you need to create below hugepages_settings.sh shell script (latest script can also be downloaded from My Oracle Support note 401749.1) for calculating the recommended Huge pages value:
#!/bin/bash
#
# hugepages_setting.sh
#
# Linux bash script to compute values for the
# recommended HugePages/HugeTLB configuration
#
# Note: This script does calculation for all shared memory
# segments available when the script is run, no matter it
# is an Oracle RDBMS shared memory segment or not.
# Check for the kernel version
KERN=`uname -r | awk -F. '{ printf("%d.%d\n",$1,$2); }'`
# Find out the HugePage size
HPG_SZ=`grep Hugepagesize /proc/meminfo | awk {'print $2'}`
# Start from 1 pages to be on the safe side and guarantee 1 free HugePage
NUM_PG=1
# Cumulative number of pages required to handle the running shared memory segments
for SEG_BYTES in `ipcs -m | awk {'print $5'} | grep "[0-9][0-9]*"`
do
MIN_PG=`echo "$SEG_BYTES/($HPG_SZ*1024)" | bc -q`
if [ $MIN_PG -gt 0 ]; then
NUM_PG=`echo "$NUM_PG+$MIN_PG+1" | bc -q`
fi
done
# Finish with results
case $KERN in
'2.4') HUGETLB_POOL=`echo "$NUM_PG*$HPG_SZ/1024" | bc -q`;
echo "Recommended setting: vm.hugetlb_pool = $HUGETLB_POOL" ;;
'2.6' | '3.8' | '3.10' | '4.1' | '4.14' | '5.15' ) echo "Recommended setting: vm.nr_hugepages = $NUM_PG" ;;
*) echo "Unrecognized kernel version $KERN. Exiting." ;;
esac
# End
- Once you create or download the latest huge pages value calculation shell script provide it an execution permission and run this script as below:
- Now configure the calculated Huge Page size value in /etc/sysctl.conf file:
- Run the below command for applying the configured Huge Page value:
sysctl -p
- Run below command to validate the applied Huge pages value and check available Huge Pages:
- You can now restart the database instance and run the below command again for validating the available Huge Pages in Linux system:
grep Huge /proc/meminfo
He is a lazy System Administrator who likes to Automate his task through Shell scripts and Ansible playbooks. He has contributed his many Automations work in Kyndryl and also maintain his own internal company Github repo.
His hobby is researching on new technologies and sharing this through his blog but if he is not researching or writing blog then you will find him listening music while sipping cup of coffee and enjoying the nature view.
You can find him on Twitter, Facebook, Linkedin or send him an email at webmaster@cjcheema.com