Special requirements apply to configuring the persistence (data and log) of an SAP HANA VM.

To achieve maximum SAP HANA scalability and performance for four HANA VMs on the P-Series models, we recommend using Linux Logical Volume Manager (LVM) devices for the SAP HANA persistence. LVM enables a striping of the logical volumes across multiple vSAN devices, which provides better performance and flexible administration. If capacity adjustments are required, LVM also enables resizing of logical volumes.

Note: Using Linux LVM is optional. Use it to achieve maximum scalability and performance. Single devices can also be created for each SAP HANA VM that match the required capacity for data and log files and are formatted for XFS.

The following sections provide step-by-step instructions for configuring the SAP HANA persistence on Linux LVM devices.

Configuring the SCSI controller

A VMware best practice is to use a dedicated SCSI controller for the SAP HANA data and log devices. Configure the SCSI controller as VMware Paravirtual. The example shown in the following figure uses SCSI Controller 0 for the system device, SCSI Controller 1 for the LVM data devices, and SCSI Controller 2 for the LVM log devices:

Figure 4.              VM SCSI controller for the SAP HANA persistence

Configuring the device capacity for data and log

Use the SAP Quick Sizer tool to determine the required capacity for the SAP HANA data and log devices.

If the sizing details are not available, configure the sizes of the data and log devices based on the memory size of the SAP HANA VM. For data, configure 2 x RAM. For log, configure 0.5-1 x RAM (512 GB maximum).

Add the capacity of the data and the log device to get the total capacity required for the SAP HANA persistence of the VM.

Assigning VM storage policies for SAP HANA data and log

Define a storage policy for the SAP HANA physical devices that are deployed on the vSAN data store. While the default vSAN storage policy is adequate for the system device, the SAP HANA persistence devices must have a dedicated storage policy to meet the SAP performance requirements. The following figure shows the rules that we defined in the storage policy for the SAP HANA persistence:

Figure 5.              VM storage policy for the SAP HANA persistence

vSAN storage policy requirements for the SAP HANA persistence

The vSAN storage policy requirements are:

• Failure tolerance method = RAID 1 (mirroring)
  This option provides best performance on vSAN compared to RAID 5/6, which is mainly optimized for space utilization.
• Failures to tolerate = 1
  This setting defines the number of disk, host, or fault-domain failures that a storage object can tolerate.
• Number of disk stripes per object = 4
  This setting defines the number of disks across which each replica of a storage object is striped.

Note: With RAID 1, each storage object has two replicas.

The optimal number depends on the number of available disk groups and capacity disks. The minimum (default) is 1 and the maximum is 12.

• Object space reservation = Thick Provisioning
  Defines that the devices will be thick-provisioned.
• Flash read cache reservation (%) = 0
  This option applies only to hybrid vSAN deployments—NVMe or SSD for cache and HDDs for capacity.
• Disable object checksum = Off
  During read/write operations, vSAN checks the validity of the data based on the checksum. If the data is not valid, vSAN takes the necessary steps to either correct the data by retrieving a new copy from another replica or report the issue to the user for action (nonrecoverable errors).
• Force provisioning = off
  If this option is ON, the object is provisioned even if the policy specified in the storage policy is not satisfiable with the resources that are available in the cluster.

Assigning vSAN devices for the SAP HANA persistence to the VM and using them in LVM

Linux LVM distributes the logical volumes that the application uses across multiple physical devices, in this case, vSAN devices. We recommend distributing the persistence for the data device across eight physical devices and distributing the persistence for the log device across four physical devices.

In the following example, we need 500 GB capacity for the SAP HANA data device and 200 GB for the SAP HANA log device, giving a total of 700 GB for the SAP HANA persistence. We therefore create eight 62 GB vSAN devices for data and four 52 GB devices for log.

Note: Using Linux LVM enhances performance but is optional. Single devices matching the required capacity can be created for data and log files and formatted for XFS for each SAP HANA VM.

Every volume uses the VM storage policy HANA PERSISTENCE and is assigned to the dedicated SCSI controller, as shown in the following figure:

Figure 6.              Assigning a VM storage policy to the devices

Adding devices to Linux LVM and creating the SAP HANA persistence

After the devices have been added to the VM, perform the following steps in Linux to partition the devices and add them to LVM:

1. Rescan the SCSI devices by running the following Linux command:

rescan-scsi-bus.sh

2. Verify that the devices are visible by running the fdisk –l command.

3. Create a Linux LVM partition on each device. The best way is to create a text file (for example, fdin.txt) with the following content:

n
p
1


t
8e
w

4. Create a partition on a device by running a command such as:

cat fdin.txt | fdisk /dev/sdb

This command creates LVM partitions on the devices, for example, /dev/sdb1. The partitions can be used as physical devices in LVM.

Repeat this command for every device that you create for the SAP HANA persistence.

5. Add the devices as physical devices to LVM:

pvcreate /dev/sdb1 /dev/sdc1 /dev/sdd1 /dev/sde1 /dev/sdf1 /dev/sdg1 /dev/sdh1 /dev/sdi1 /dev/sdj1 /dev/sdk1 /dev/sdl1 /dev/sdm1

6. Create an LVM volume group for data with the name hana-data:

vgcreate hana-data /dev/sdb1 /dev/sdc1 /dev/sdd1 /dev/sde1 /dev/sdf1 /dev/sdg1 /dev/sdh1 /dev/sdi1

7. Create a second LVM volume group for log with the name hana-log:

vgcreate hana-log /dev/sdj1 /dev/sdk1 /dev/sdl1 /dev/sdm1

8. Create logical volumes for the SAP HANA data and log file systems on the new volume groups using lvcreate. For example:

lvcreate --name data --size 500G --stripes 8 --stripesize 64 hana-data

and

lvcreate --name log --size 200G --stripes 4 --stripesize 64 hana-log

9. Use the number of physical devices that you created earlier as the --stripes parameter.
10. Format the devices by running:

mkfs.xfs /dev/mapper/hana--data-data and mkfs.xfs /dev/mapper/han--log-log

11. List the formatted devices under the required mount points for SAP HANA by running (for example):

mount /dev/mapper/hana-data /hana/data and mount /dev/mapper/hana-log /hana/log

12. To persist the mounts over reboots, add the corresponding entries to /etc/fstab.