Best Practice 1.2 – Implement infrastructure monitoring for SAP

Set up your infrastructure monitoring to provide information about supporting services that are used to keep your SAP application running and supporting your users. Some examples include CPU and memory utilization, storage and filesystem usage and performance (IOPS and throughput), and network throughput. Include any dependent foundational services used by SAP, such as on-premises Active Directory services, DNS, and third-party tools, such as high availability (HA) and backup software. Evaluate AWS tools and SAP-specific tools from the AWS Marketplace that can help correlate and visualize this information, such as DataDog, Splunk, DynaTrace, and Avantra. Use this information to identify trends and determine when a corrective action is required.

Suggestion 1.2.1 - Implement CloudWatch metrics and alarms for services supporting SAP

Implement Amazon CloudWatch detailed monitoring metrics and thresholds with alarms for all of your SAP systems. These metrics and alarms should include monitoring for common problems which can affect SAP system availability and performance. Common infrastructure monitoring areas focus on Amazon Elastic Compute Cloud (EC2) instances, Amazon Elastic Block Storage (Amazon EBS) volumes, and Elastic Load Balancing (ELB).

Common monitoring items include the following:

Base your alarm thresholds on healthy patterns of historical production usage of your system. Continually review and tweak your alarm thresholds to prevent problems.

Review the following resources to get started:

Suggestion 1.2.2 - Implement AWS service quota monitoring for SAP services

Implement a monitoring tool, such as Amazon CloudWatch, or other process to keep track of your AWS service quotas for required SAP resources in your landscape. Amazon EBS and Amazon EC2 instance quotas are the most common quotas to affect a SAP workload.

For EC2 instance quotas, consider that SAP landscapes can often use a mix of Amazon EC2 instance types and that some types have a different On-Demand service quota. For example, the x* and u* (High Memory) EC2 instance types have a different service quota that is separate from the combined quota for standard types like c6, m6, and r6 instance families.

When planning new or scaling existing workloads, verify that your service quotas will support this and engage Support if a quota increase is required.