Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Calculate EBS Storage Requirements for SAP HANA

Overview

This guide provides storage configuration recommendations for SAP HANA workloads running on Amazon EC2. Learn how to configure Amazon EBS volumes to meet SAP’s storage key performance indicators (KPIs).

SAP HANA stores and processes its data primarily in memory, and provides protection against data loss by saving the data to persistent storage locations. To achieve optimal performance, the storage solution used for SAP HANA data and log volumes should meet SAP’s storage KPI requirements. Amazon has worked with SAP to certify both Amazon EBS General Purpose SSD (gp2 and gp3) and Provisioned IOPS SSD (io1, io2 Block Express) storage solutions for SAP HANA workloads.

New Amazon EBS Storage Guidelines for SAP HANA

This document introduces a memory-based formula approach for storage sizing, replacing previous instance-specific recommendations. This change enables customers to better understand storage configuration logic and maintain greater control over performance optimization decisions.

The new guidance focuses on gp3 and io2 Block Express volumes as the current standard recommendation for all new deployments. While gp2 and io1 volumes remain supported for existing deployments, we recommend gp3 for new implementations due to its predictable performance and cost-effectiveness, with io2 Block Express as the upgrade path for systems requiring additional performance.

Testing with SAP HANA Hardware and Cloud Measurement Tools

Amazon has ensured that the storage configuration guidelines meet the Key Performance Indicators (KPIs) for running SAP HANA. However, for workloads with high performance requirements or those that deviate from the standard recommendation, we strongly recommended validating the performance of your storage configuration with SAP HANA Hardware and Cloud Measurement Tools.

See:

EBS Storage Volume Configurations

This section provides the formulas and methodology for calculating SAP HANA storage requirements. The calculations factor in memory size and workload characteristics to determine appropriate volume sizing and performance configuration. Adjust these baseline recommendations based on your specific workload requirements and growth projections.

Refer to SAP HANA EBS Storage Reference for calculated requirements based on available memory.

Root and SAP Binary Volumes

The root volume contains operating system files, configuration, and logs. The sizing recommendation is suitable for most SAP HANA deployments and sizes, but may vary based on your AMI strategy and non-SAP software requirements. Consider using a separate volume for the /tmp filesystem in high-usage environments.

The SAP binaries volume (default /usr/sap/<SID>/) contains common SAP executables and binaries.

Storage Type

Use gp3 volumes for root storage. The baseline performance characteristics meet operating system requirements.

Size Calculation

root_and_sap_volume_size = 50 GiB

IOPS Formula

root_and_sap_iops_target = 3000 (fixed)

Throughput Formula

root_and_sap_throughput_target = 125 MB/s (fixed)

Example

ANY Memory System Root Volume:

HANA Data Volume

The Storage\DATA filesystem (default /hana/data) stores the persistent copy of the SAP HANA in-memory database. While SAP HANA operates with data in-memory, this volume ensures data durability through regular savepoints. The storage must handle mixed workload patterns including random reads and writes during normal operation, and sequential patterns during savepoints, with consistently low latency to maintain database performance.

Size Calculation

The data volume size requirements are derived from system memory size. While actual storage requirements depend on your specific workload, compression ratios, and growth projections, use the following calculation as a baseline. Consult SAP sizing tools for precise calculations.

data_volume_size = MROUND(memory_size * 1.2, 100)

Where:
- Size factor = 1.2
- Rounding factor = 100

IOPS Formula

data_iops_target = MROUND(7200 + (0.45 * memory_size), 100)

Where:
- Base IOPS = 7200
- IOPS factor = 0.45 per GiB of memory
- Rounding factor = 100

Throughput Formula

data_throughput_target = MIN(MROUND(450 + (0.2 * memory_size), 125), 2000)

Where:
- Base throughput = 450 MB/s
- Throughput factor = 0.2 MB/s per GiB of memory
- Maximum throughput = 2000 MB/s (see exception)
- Rounding factor = 125

Volume Striping

Implement volume striping when you need to meet specific technical limits, performance requirements, or operational demands. Refer to the When to Stripe Volumes for detailed guidance on when striping is appropriate.

For gp3 volumes, throughput is typically the first limit you’ll encounter. For io2 Block Express volumes, throughput is calculated as IOPS × I/O size. SAP HANA workloads typically use 256 KiB I/O operations - at this size, a single io2 Block Express volume can achieve 4,000 MB/s throughput with 16,000 IOPS. Given these capabilities, volume striping is not required for most HANA deployments on io2 Block Express. If higher throughput is needed, you can adjust the provisioned IOPS accordingly.

If implementing striping for data volumes, use a 256 KB stripe size to optimize for data operations.

Examples

512 GiB Memory System HANA Data Volume:

4 TiB Memory System HANA Data Volume:

HANA Log Volume

The Storage\LOG filesystem (default /hana/log) stores the redo log files that ensure data durability and consistency. This filesystem handles write-intensive workloads with high-frequency, small, sequential writes. Because log writes directly impact database response time and transaction performance, storage volumes require consistent sub-millisecond latency.

Size Calculation

The log volume size requirements are derived from system memory size. Modifications can be made based on transaction volume and log backup frequency.

log_volume_size = MROUND((memory_size * 0.5),100)

Where:
- Minimum Size = 50 GiB
- Maximum Size = 500 GiB
- Rounding factor = 100

IOPS Formula

log_iops_target = 3000

Where:
- Base IOPS = 3000

Throughput Formula

log_throughput_target = MIN(MROUND(300 + (0.015 * memory_size), 300), 500)

Where:
- Base throughput = 300 MB/s
- Throughput factor = 0.015 MB/s per GiB of memory
- Maximum throughput = 500 MB/s
- Rounding factor = 300

Volume Striping

For log volumes, striping is generally not required to achieve the log_throughput_target when using gp3 or io2 Block Express volumes. Single volumes typically provide sufficient performance for log operations.

If implementing striping for log volumes, use a 64 KB stripe size to optimize for sequential write patterns typical of log operations. Refer to the When to Stripe Volumes section to understand where striping is required in order to achieve the throughput, IOPS or performance targets.

Examples

512 GiB Memory System HANA Log Volume:

HANA Shared Volume

The HANA Shared filesystem (default /hana/shared) contains SAP HANA installation files, trace files, and shared configuration files.

Size Calculation

For single-node deployments:

shared_volume_size = MIN(memory_size, 1024)

Where:
- memory_size is system memory in GiB
- 1024 represents 1 TiB maximum

For scale-out deployments:

For scale-out SAP HANA systems, the /hana/shared filesystem requires disk space equal to one worker node’s memory for every four worker nodes in the deployment.

shared_volume_size = worker_node_memory * CEILING(number_of_worker_nodes/4)

Where:
- worker_node_memory is the memory size of a single worker node in GiB
- number_of_worker_nodes is the total number of worker nodes
- CEILING rounds up to the nearest whole number

Examples for scale-out deployments

IOPS Formula

shared_iops_target = 3000

Where:
- Base IOPS = 3000 (fixed)

Throughput Formula

shared_throughput_target = 125

Where:
- Base throughput = 125 MB/s (fixed)

Examples

512 GiB Memory System HANA Shared Volume:

HANA Backup Volume (Optional)

The /backup filesystem provides local storage for SAP HANA file-based backups, including data and log backups. While local filesystem backups can be useful for non-critical systems or as a secondary backup option, they present several challenges in production environments:

Size Calculation

The size of the backup volume is very dependent on the usage of the system. The following should be used as an initial baseline, but adapted post deployment depending on backup size, volume of change, retention of local copies and contingency.

backup_volume_size = memory_size * 3

Where:
- memory_size is system memory in GiB

IOPS Formula

backup_iops_target = n/a

Throughput Formula

For ST1 volumes, use this formula as a starting point to determine the number of volumes needed for backup throughput. Adjust the final volume count based on your actual backup window requirements and performance monitoring data.

backup_volumes_for_throughput = CEILING(memory_size/6000) * 500

Where:
- memory_size is system memory in GiB
- 6000 represents the GiB threshold for striping
- 500 is maximum throughput MB/s per st1 volume
- Result indicates number of volumes needed for throughput

Examples for scale-out deployments

When to Stripe Volumes

Linux Logical Volume Management (LVM) striping distributes data across multiple EBS volumes to increase I/O performance. The striped volumes act as a single logical volume, with reads and writes distributed across all volumes in the stripe set.

Implement storage volume striping in the following scenarios: