Right Sizing AWS EC2 Instances for SAP Application Servers

When performing right-sizing for SAP workload on AWS, determining the appropriate sizing for database servers is generally less complicated as we would have access to database growth and performance numbers, and end-user pain points are known. In the case of the SAP HANA database, the process is even relatively simpler due to the availability of a specific set of SAP-certified EC2 instances to choose from and the choice would be based on the assessment and analysis conducted using the SAP S/4HANA sizing report and other relevant documents.

However, right-sizing for SAP Application instances e.g. ASCS, ERS, and Additional Application Instances (AAS) can present some challenges as direct information and requirements are not readily available. We may encounter questions such as determining the optimal number of additional application instances required to meet load balancing needs for the production system or identifying the appropriate compute capacity of application servers to meet the performance requirements of the production workload.

In this blog, I will discuss a step-by-step approach that I have used for the creation of SAP application instances sizing documentation and TCO preparation on AWS for a migration scenario, leveraging data from various sources, including the latest Configuration Management Database (CMDB), most recent SAP EarlyWatch Alert (EWA) reports, on-premises server performance reports, SAPS data, user questionnaires, and finally the AWS Migration Portfolio Assessment (MPA) tool for the recommendation. I am deliberately leaving out any database server sizing topic in this blog.

I. Understanding the SAP Application Architecture

To make informed decisions regarding EC2 instance sizing, it is crucial to understand the SAP application architecture and the role of different application instances. This section briefly introduces the ASCS, ERS, and AAS components, their functions, and their interdependencies within an SAP landscape.

• SAP ASCS Instance: The SAP ASCS (Application Server Central Services) instance is a key component in SAP system architecture, managing locks, facilitating communication, and serving as a central instance for administration. It ensures data consistency, high availability, and coordination within the SAP landscape. The tasks performed by the ASCS instance are compute-intensive hence the right instance with optimal compute power is recommended.
• SAP ERS Instance: The SAP ERS (Enqueue Replication Server) instance is a specialized component in an SAP system landscape that replicates lock information from the primary ASCS instance to provide high availability. It ensures that lock data remains consistent across multiple ASCS instances, allowing for failover and uninterrupted operations in case of a primary instance failure. Generally would follow the same sizing guideline as an ASCS instance.
• SAP AAS Instance: The SAP Additional Application Server (AAS) instance (Dialog Instance) is an additional server within an SAP system landscape that runs application processes. It helps distribute the workload, enhances system performance, and allows for scalability. Multiple AAS instances can be added to accommodate increased user demand and optimize system resource utilization. Hence these instances are most critical from the SAP application server sizing perspective.

II. Gathering On-Premises Data

A crucial step in right-sizing SAP application instances is collecting relevant data from the on-premises environment. This includes:

• CMDB Data: Extracting the configuration details of existing on-premises servers, including CPU, memory, storage, and network specifications.
• SAP EWA Report: Analyzing the EarlyWatch Alert (EWA) reports provided by SAP to understand system performance, utilization, and potential bottlenecks or areas where the system may be over or underutilized.
• On-Premises Server Performance Report: Examining detailed performance reports to assess CPU, memory, and I/O utilization, identifying peak usage periods, and determining baseline performance metrics to ensure sufficient capacity on AWS EC2 instances.
• On-Premises Server SAPS Data: Evaluating SAPS (SAP Application Performance Standard) values, which measure the processing power required by an SAP system, and understanding the SAPS-to-CPU ratio. SAPS data provides a baseline for comparing and sizing the compute capacity required in the AWS environment.

III. Conducting a Requirements Assessment

To determine the appropriate size of AWS EC2 instances, it is essential to assess the organization's requirements. This can be achieved through the following steps:

• Questionnaire: Designing a comprehensive questionnaire to gather information about the current and future requirements, expected user load, peak usage periods, growth projections, and any specific performance concerns.
• Performance Analysis: Analyzing the collected on-premises data, identifying performance bottlenecks, and understanding the resource requirements for each SAP application instance.
• Scalability and Availability Needs: Assessing the organization's scalability and availability requirements to ensure the chosen EC2 instance types can handle the workload growth and provide the necessary resilience.

IV. Leveraging AWS Migration Portfolio Assessment (MPA) Tool

AWS provides the Migration Portfolio Assessment (MPA) tool to the APN partner to simplify the assessment process of the on-premises portfolio in preparation for the migration to AWS. It automates the analysis of on-premises applications and infrastructure. This section explores how to utilize MPA for EC2 instance sizing for SAP application instances:

• Data Collection and Analysis: Upload the collected on-premises data, including CMDB data, performance reports, SAPS data, and questionnaire responses, to the MPA tool for analysis. Data can be uploaded in CSV or an Excel formatted file.

• Sizing Recommendations**: Review the MPA tool's output, as it provides insights into instance type recommendations, such as CPU, memory, storage, and network capacity, based on the workload characteristics and requirements.

** There are no specific sizing guidelines for SAP servers recommendation in the MPA tool. To get results relevant to SAP-certified and supported EC2 instances I customize AWS server recommendation under the "Modify Assumptions" section to exclude non-supported EC2 instances from consideration.

• Cost Analysis: MPA tool considers the cost implications of the recommended EC2 instance types, including instance pricing, storage options, and licensing considerations, to achieve a cost-effective SAP deployment on AWS. Provides EC2 BOM (Bill of Material) analysis and TCO comparison with an equivalent on-premises hosting.

V. Instance Sizing Best Practices and Considerations

This section covers essential best practices and considerations when choosing the right size AWS EC2 instances for SAP application instances:

• Instance Families and Types: Understanding the various EC2 instance families and types available on AWS, such as general purpose, memory-optimized, and compute-optimized, and selecting the most appropriate options for SAP workloads (refer to SAP note 1656099 for a list of SAP-supported EC2 instances).
• Resilience and High Availability: Ensuring the chosen EC2 instances provide the necessary resilience and high availability for critical SAP application instances, taking advantage of features like Auto Scaling (Where applicable), Multi-AZ deployments, and load balancing (refer to SAP on AWS High Availability documentation)
• SAP Note and Documentation: Review SAP notes and documentation specific to AWS deployments to understand any platform-specific considerations or limitations (refer to SAP notes 1656099 and 1656250).

VI. Conclusion

Choosing the right size AWS EC2 instances for SAP application instances is a critical aspect of a successful migration. By leveraging various data sources, including the ones explained in this blog, organizations can make informed decisions to optimize their SAP workload on AWS. The approach outlined in this blog has helped me in defining an optimal methodology to ensure the successful migration and right-sizing of SAP application instances on AWS, enabling customers to reap the benefits of cloud scalability, flexibility, and cost optimization.