Best practices for SAP performance benchmarking

Performance benchmarking is an essential part of implementing any SAP system. By benchmarking your system at different stages of the implementation process, you can identify and address any performance bottlenecks early on. This can help you ensure that your SAP system is able to meet the needs of your business as it grows without any performance impact.

Performance benchmarking means different things at different stages of a project.

• Pre-implementation: At this stage, you should benchmark the performance of your SAP system on your on-premises environment. This will give you an indication of how to size your cloud resources along with providing a baseline to compare against once you’ve migrated your SAP system to cloud.
• Implementation: Once you’ve deployed your SAP system to Google Cloud, you should benchmark the performance of your system again. This will help you identify any performance bottlenecks that may have been introduced during the implementation process. In this phase, perform different types of testing such as load testing and performance testing to identify any potential performance bottlenecks. Load testing can be performed using a variety of tools and methods. One common approach is to use a third-party tool such as Load Runner or Jmeter to simulate a large number of users accessing the system. Synthetic transactions can also be used to simulate the execution of specific business processes. When defining load test scripts, it’s important to consider the types of users, business processes, and peak load that the system is expected to experience.
• Operations: Once your system is migrated to Google Cloud, it’s important to ensure that it’s meeting your performance needs. This will help you identify any performance degradation that may occur over time and will help you identify any areas where your system is underperforming so you can make necessary improvements.

Before migrating your SAP system to the cloud, consider the following to improve performance:

1. Network

Network tier

Google Cloud offers two Network Service tiers: Standard and Premium. Using Network Service Tiers allows customers to improve network performance. A key component that sets Google Cloud apart is its premium network backbone that offers high network performance and low latency.

In Premium Tier, traffic between the internet and VM instances in your VPC network is routed by keeping it within Google's network as much as possible, thus providing the highest performance and low latency. In Standard tier, traffic between the internet and VM instances in your VPC network is routed over the internet in general.

Latency from on-premises to primary and DR region

When selecting a region for your SAP systems, consider the latency between your on-premises environment and the target primary and DR region in the cloud. In Google Cloud, you can use the Performance Dashboard to check the latency between different Google Cloud regions.

Below are different scenarios where latency can have an impact:

• High latency between an application and a database can cause slow response times, increased transaction times and timeout errors.
• Low zone-to-zone latency is important for high availability because SAP systems can be configured to replicate data across zones. However, if the zone-to-zone latency is too high, it can impact database replication, resulting in a performance impact on the SAP system.
• Low region-to-region latency is important to achieve your disaster recovery recovery point objective (RPO).

Interconnect and Cloud VPN

Google Cloud offers a variety of Interconnect options, including Dedicated Interconnect, Cloud Interconnect, and Cloud VPN.

The best interconnect option for your SAP system will depend on your specific needs. For high-bandwidth and low-latency requirements, Dedicated Interconnect or Partner Connect are good options. If you already have a partner connection to another cloud provider, you can reuse it to connect to Google Cloud if the partner supports it. If you require a more cost-effective solution, Cloud HA-VPN may be a better option and it will suffice for most SAP applications. However, latency can vary, as last-mile traffic is routed over the internet.

2. Availability zone

When selecting a zone for your SAP system, it is important to consider the following factors:

• Latency: The latency between different zones in case of a multi-zone architecture
• Availability: The availability of the machine types in the selected zones

You can use Google Cloud's Performance Dashboard to check the latency between different zones.

3. Disk selection

Once you’ve selected a zone, you need to select type and size the disk as per the use cases and required IOPS and throughput.

Compute Engine offers different types of persistent disks and Hyperdisks based on either solid-state drive (SSD) or standard hard disk drive (HDD) technology. Each type has different performance characteristics. Google Cloud manages the underlying hardware of the disks to ensure data redundancy and to optimize performance.

Disk throughput is the amount of data that can be read from or written to disk per second. It’s important to ensure that your disk throughput and IOPS are sufficient to meet the needs of your SAP system.

You can refer to SAP Note - 2456406 for more information on the type of disk supported for different SAP products. To achieve your required IOPS and throughput, size the selected disk to meet the required performance as per the documentation.

To benchmark the disk performance, you can use Flexible I/O tester (FIO) tool.
You can learn more in this document.

To measure IOPS and throughput of a disk in use on a running instance, benchmark the file system with its intended configuration. Use this option to test a realistic workload without losing the content of your existing disk. Note that when you benchmark the file system on an existing disk, there are many factors specific to your development environment that may affect benchmarking results, and you may not reach the disk performance limits.

To measure the raw performance of a persistent disk, benchmark the block device directly. Use this option to compare raw disk performance to disk performance limits.

4. VM instance selection

In Google Cloud, you can deploy SAP systems on certified predefined machine type or a custom machine type that has a SAPS rating that is greater than or equal to the SAPS rating of your SAP instance. You should also ensure that the VM instance has enough vCPUs and memory to meet the requirements of your SAP instance. Refer to SAP Note - 2456432 and Google Cloud documentation for list of SAP certified machine types.

Persistent disk performance scales with the size of the disk and with the number of vCPUs on your VM instance. Performance scales until it reaches either the limits of the disk or the limits of the VM to which the disk is attached. The machine type and the number of vCPUs on the VM determine the VM limits.

Example

Consider a 1,000 GB zonal SSD persistent disk attached to a VM with an N2 machine type and 4 vCPUs. The read limit based solely on the size of the disk is 36,000 IOPS (6,000 baseline IOPs + (30 IOPS per GB * 1,000 GB)). However, the VM has 8 vCPUs, so the read limit is restricted to 15,000 IOPS.

5. Shared file systems

Shared file systems are a critical component of many SAP systems. They can be used to provide storage for SAP application files, interface files and configuration files.

It is important to benchmark share volumes to ensure that they meet the required performance for SAP workloads. SAP workloads can be very demanding, and share volumes that are not performing well can lead to performance bottlenecks for the entire SAP system.

There are a number of factors that can affect the performance of shared file system, including:

• Network bandwidth: The bandwidth of the network that connects the shared file system to the SAP servers can have a significant impact on performance.
• Storage performance: The performance of the storage devices that are used for shared file systems can also have a significant impact on performance.

Summary

Benchmark your SAP system at different stages of implementation to identify and address performance bottlenecks early on and throughout the system’s lifecycle. This will ensure that your SAP system can meet the growing performance needs of your business.

Consider all aspects of the underlying infrastructure, including network tier, interconnect option, latency, availability zone, disk selection, VM instances, and shared file systems, when benchmarking. To learn more, visit SAP on Google Cloud.