IBM Cloud Docs
Bare Metal Server certified profiles on VPC infrastructure for SAP HANA

Bare Metal Server certified profiles on VPC infrastructure for SAP HANA

Profiles list

The following table gives you an overview of the SAP-certified profiles with bare metal servers for VPC. The vCPUs in this list are CPU cores and their secondary threads. The term vCPU is kept for comparison with their virtual counterparts.

Table 1. IBM Cloud Bare Metal Servers for VPC certified for SAP HANA
Profile vCPU Memory (RAM GiB) SAPS SAP HANA
Processing Type
Compute Optimized
cx2-metal-96x192 96 192 107,400 SAP Business One (**)
Balanced
bx2d-metal-96x384 96 384 124,130 OLTP/OLAP (*)
SAP Business One (**)
Memory Optimized
mx2d-metal-96x768 96 768 127,620 OLTP/OLAP (*)
SAP Business One (**)
Ultra High Memory Optimized
ux2d-metal-112x3072 112 3,072 140,730 OLTP/OLAP (*)
ux2d-metal-224x6144 224 6,144 294,730 OLTP/OLAP (*)

(*): RHEL 8.4 for SAP Solutions, RHEL 8.6 for SAP Solutions, RHEL 8.8 for SAP Solutions, RHEL 9.0 for SAP Solutions, RHEL 9.2 for SAP Solutions
SLES 15 SP3, SLES 15 SP4, SLES 15 SP5

(**): SLES 15 SP3, SLES 15 SP4

For more information, see SAP Note 2927211 - SAP Applications on IBM Cloud Virtual Private Cloud (VPC) Infrastructure environment.

For SAP HANA deployments that use IBM Cloud Bare Metal Servers for VPC, only single-node deployments are supported. Multi-node / scale-out is not currently supported.

Understanding Bare Metal Server profile names

With IBM Cloud Bare Metal Servers for VPC, the profile families that are certified for SAP are: Compute Optimized, Balanced, Memory Optimized, and Ultra High Memory Optimized.

  • Compute Optimized family profiles provide more compute power, and they have more cores with less memory.
  • Balanced family profiles provide a good mix of performance and scalability for more common workloads.
  • Memory Optimized and Ultra High Memory Optimized family profiles cater to memory intensive workloads, such as demanding database applications and in-memory analytics workloads, and are especially designed for SAP HANA workloads.

For more information, see x86-64 bare metal server profiles.

The first letter of the profile name indicates the profile family:

Table 2. IBM Cloud Bare Metal Servers for VPC Profile Families
First letter Characteristics of the related profile family
c Compute Optimized family, vCPU to memory ratio 1:2
b Balanced family, vCPU to memory ratio 1:4
m Memory Optimized family, higher vCPU to memory ratio 1:8
u Ultra High Memory Optimized family, even higher vCPU to memory ratio 1:27.43

The bare metal server profile names are contextual and sequential. See the following example:
Table 3. Profile naming for SAP HANA
Profile name Naming convention component What it means
mx2d-metal-96x768 m Memory Optimized family
x Intel x86_64 CPU Architecture
2 The generation for the underlying hardware
d the optional 'd' in the name indicates that the server is equipped with one or more SSD storage devices
spacer
metal metal in the name indicates that this is a bare metal server
spacer
96 96 vCPU
x spacer
768 768 GiB RAM

Profiles available on Hourly Consumption Billing

All IBM Cloud Bare Metal Servers for VPC are available with Hourly Consumption Billing, which includes Suspend Discounts and Sustained Usage Discounts. With Suspend Discounts, storage charges occur only if the server is in shutdown state. With Sustained Usage Discount, the more a server is used, the less the cost per hour.

Storage specifications

When the bare metal server profiles for SAP HANA are initially provisioned, the servers all have one pre-configured disk (sda) attached with the following basic layout:

Table 3. Storage configuration of the default bare metal server deployment (boot volume)
File system Partition Storage type Size
sda1 Pre-configured BIOS volume 1 MB
/boot/efi sda2 Pre-configured boot volume 100 MB
/ sda3 Pre-configured root volume 9.9 GB

In addition to these partitions, bare metal servers for VPC have 8 - 16 NVMEs – depending on their size – which need to be configured after the server deployment.

To fulfill the KPIs defined for SAP HANA, each profile needs different storage volumes that are listed in detail in the following sections. These storage configurations are mandatory, not sample storage configurations, because they are the tested and certified storage layouts that comply with SAP HANA Tailored Data Center Integration (TDI) Phase 5. The recommendation is to use these specifications.

Customers who want to choose different layouts are advised to follow the SAP HANA TDI Overview and SAP HANA TDI FAQ when they configure different layouts. In that case, users must run SAP's performance measurement tool HCMT - see SAP Note 2493172 - SAP HANA Hardware and Cloud Measurement Tools and follow the instructions of the HCMT guide to check compliance with SAP’s KPIs.

This holds true especially, if file shares are used for SAP HANA installations. They can be deployed and mounted in arbitrary ways to provide additions storage, for example for backups, as needed. For SAP HANA data and log files, however, they have to be evaluated.

In any case, SAP's recommended file system layout must be available for SAP HANA deployment.

Bare metal servers for VPC - Storage Layouts

The following table shows the required physical volumes, related volume groups, logical volumes, and their characteristics:

Table 4. Storage layout for Bare metal servers for VPC
Profile File
system
Logical
Volume
LV Size
(GB)
Volume Group Physical
Volume
PV Size
(TB)
cx2d-metal-96x192 /hana/shared hana_shared_lv 192 vg0 nvme0n1-
nvme3n1-
11.6
/hana/data hana_data_lv min. 576 vg1 nvme4n1-
nvme7n1-
11.6
/hana/log 192 vg0







bx2d-metal-96x384 /hana/shared hana_shared_lv 384 vg0 nvme0n1-
nvme3n1-
11.6
/hana/data hana_data_lv min. 1,152 vg1 nvme4n1-
nvme7n1-
11.6
/hana/log 384 vg0







mx2d-metal-96x768 /hana/shared hana_shared_lv 768 vg0 nvme0n1-
nvme3n1-
11.6
/hana/data hana_data_lv min. 2,304 vg1 nvme4n1-
nvme7n1-
11.6
/hana/log 512 vg0







ux2d-metal-112x3072 /hana/shared hana_shared_lv 3,072 vg0 nvme0n1-
nvme3n1-
11.6
/hana/data hana_data_lv min. 9,216 vg1 nvme4n1-
nvme7n1-
11.6
/hana/log 512 vg0







ux2d-metal-224x6144 /hana/shared hana_shared_lv 6,144 vg0 nvme0n1-
nvme1n1-
25.6
/hana/data hana_data_lv min. 18,432 vg0
/hana/log 512 vg0


Profile ux2d-metal-224x6144 is equipped with different set od disks, so jump directly to "Steps for setting up storage for the ux2d-metal-224x6144 profile".

Steps for setting up storage for the profiles up to 3,072 GiB

Note, that both volume groups vg0 and vg1 are not fully used. Remaining space can be used to extend the listed sizes, which are only minimum sizes, or can be used for other purposes. However, do not point I/O load at the remaining space since that impacts SAP HANA’s performance.

To ensure a higher level of availability and failure resilience, RAID10 logical volumes are built on-top the under-laying NVMEs, based on Linux’s logical volume manager.

These steps show a step-by-step guide for setting up the volume groups, logical volumes, and file systems. Size information differs and can be retrieved from the storage layout table.

  1. Log in to the OS and install the lvm2 package, if not installed already.

    [root@mx2d-metal-96x768 ~]# yum install lvm2
    

    This command applies to RHEL, on SLES use ‘zypper install’ instead.

  2. Create the volume groups.

    [root@mx2d-metal-96x768 ~]# vgcreate vg0 /dev/nvme0n1 /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1
    [root@mx2d-metal-96x768 ~]# vgcreate vg1 /dev/nvme4n1 /dev/nvme5n1 /dev/nvme6n1 /dev/nvme7n1
    
  3. Create logical volumes on top of the volume groups.

    [root@mx2d-metal-96x768 ~]# lvcreate --type raid10 -i 2 -m 1 -L 768G -I 64 -n hana_shared_lv vg0
    

    768G needs to be adapted to the hana_shared volume size specific to your memory size in the sizing table. Sizes for hana_log can be found there, too, and can be used in the lvcreate command as well.

    [root@mx2d-metal-96x768 ~]# lvcreate --type raid10 -i 2 -m 1 -L 512G -I 64 -n hana_log_lv vg0
    
  4. Create hana_data. Either modify the size with the -L option according to the sizing table, or the use the entire volume group with -l 100%FREE:

    [root@mx2d-metal-96x768 ~]# lvcreate --type raid10 -i 2 -m 1 -l 100%FREE -I 64 -n hana_data_lv vg1
    
  5. Create file systems on the logical volumes. In this example, XFS is used and is mounted by label. Mount by label is not a requirement and can be adapted according to your needs:

    [root@mx2d-metal-96x768 ~]# mkfs.xfs -L HANA_SHARED -K /dev/mapper/vg0-hana_shared_lv
    [root@mx2d-metal-96x768 ~]# mkfs.xfs -L HANA_LOG -K /dev/mapper/vg0-hana_log_lv
    [root@mx2d-metal-96x768 ~]# mkfs.xfs -L HANA_DATA -K /dev/mapper/vg1-hana_data_lv
    
  6. Add the following lines to /etc/fstab and create the required directory paths with mkdir.

    LABEL=HANA_SHARED /hana/shared xfs defaults 0 0
    LABEL=HANA_LOG /hana/log xfs defaults,swalloc,inode64 0 0
    LABEL=HANA_DATA /hana/data xfs defaults,largeio,swalloc,inode64 0 0
    
  7. You can now mount the file systems.

Steps for setting up storage for the ux2d-metal-224x6144 profile

To ensure a higher level of availability and failure resilience, RAID1 logical volumes are built on-top the under-laying NVMEs, based on Linux’s logical volume manager.

These steps show a step-by-step guide for setting up the volume groups, logical volumes, and file systems.

  1. Log in to the OS and install the lvm2 package, if not installed already.

    [root@ux2d-metal-224x6144 ~]# yum install lvm2
    

    This command applies to RHEL, on SLES use ‘zypper install’ instead.

  2. Create the volume group.

    [root@ux2d-metal-224x6144 ~]# vgcreate vg0 /dev/nvme0n1 /dev/nvme1n1
    
  3. Create logical volumes hana_shared_lv and hana_log_lv on top of the volume group.

    [root@ux2d-metal-224x6144 ~]# lvcreate --type raid1 -i 2 -m 1 -L 6144G -I 64 -n hana_shared_lv vg0
    [root@ux2d-metal-224x6144 ~]# lvcreate --type raid1 -i 2 -m 1 -L 512G -I 64 -n hana_log_lv vg0
    
  4. Create logical volume hana_data_lv.

    [root@ux2d-metal-224x6144 ~]# lvcreate --type raid1 -i 2 -m 1 -l 100%FREE -I 64 -n hana_data_lv vg0
    
  5. Create file systems on the logical volumes. In this example, XFS is used and is mounted by label. Mount by label is not a requirement and can be adapted according to your needs:

    [root@ux2d-metal-224x6144 ~]# mkfs.xfs -L HANA_SHARED -K /dev/mapper/vg0-hana_shared_lv
    [root@ux2d-metal-224x6144 ~]# mkfs.xfs -L HANA_LOG -K /dev/mapper/vg0-hana_log_lv
    [root@ux2d-metal-224x6144 ~]# mkfs.xfs -L HANA_DATA -K /dev/mapper/vg0-hana_data_lv
    
  6. Add the following lines to /etc/fstab and create the required directory paths with mkdir.

    LABEL=HANA_SHARED /hana/shared xfs defaults 0 0
    LABEL=HANA_LOG /hana/log xfs defaults,swalloc,inode64 0 0
    LABEL=HANA_DATA /hana/data xfs defaults,largeio,swalloc,inode64 0 0
    
  7. You can now mount the file systems.

Check SAP Note 2777782 for RHEL and SAP Note 2684254 for SLES to adapt your OS configuration settings according to the requirements for SAP HANA.