Google Cloud's Agent for SAP planning guide

For support and monitoring, Google Cloud provides the Agent for SAP, for SAP workloads running on Compute Engine instances and Bare Metal Solution servers.

As mandated by SAP, to get support from SAP and to enable SAP to meet its service-level agreements (SLAs), you must install Google Cloud's Agent for SAP on all Compute Engine instances and Bare Metal Solution servers that run any SAP system. For more information on the support prerequisites, see SAP Note 2456406 - SAP on Google Cloud Platform: Support Prerequisites .

Version 3.6 (latest) of Google Cloud's Agent for SAP is the successor of Google Cloud's monitoring agent for SAP NetWeaver version 2, monitoring agent for SAP HANA version 2, and the Cloud Storage Backint agent for SAP HANA. Therefore, in addition to the collection of metrics, version 3.6 (latest) of Google Cloud's Agent for SAP includes the optional feature: Backint based backup and recovery for SAP HANA. You can opt in to these features that enable products and services such as Workload Manager for your SAP workloads.

For SAP systems running on a Bare Metal Solution server, Google Cloud's Agent for SAP is required when you run SAP NetWeaver connected to any of the following databases:

  • SAP HANA
  • SAP Adaptive Server Enterprise (ASE)
  • SAP MaxDB
  • IBM Db2

You install Google Cloud's Agent for SAP on the host alongside the SAP system. For instructions about how to install and configure the agent, validate your installation, and verify that the agent is running as expected, see:

If you use the following RHEL or SLES "for SAP" OS images that Google Cloud provides, then Google Cloud's Agent for SAP is packaged with the OS images:

  • RHEL: all "for SAP" images
  • SLES: SLES 15 SP4 for SAP and later versions

Supported features

Version 3.6 (latest) of Google Cloud's Agent for SAP supports the following features:

SAP Host Agent metrics collection

The SAP Host Agent metrics collection is enabled by default for agent instances installed on Compute Engine instances or Bare Metal Solution servers. Google Cloud's Agent for SAP collects and sends all the required information that SAP mandates to the SAP Host Agent, including metrics about the following:

  • CPUs, for example, CPU utilization.
  • Persistent disk storage, for example, disk throughput and latency.
  • Memory, for example, memory consumption.
  • Networks, for example, network bandwidth.
  • Configuration, including information about the host machine and its environment.

The metrics collected by Google Cloud's Agent for SAP are determined by SAP. For a description of metrics that the agent collects, see SAP Note 2469354 - Key Monitoring Metrics for SAP on IaaS Infrastructure .

SAP system discovery

This feature is at the core of the agent. It collects information related to your SAP system and the Google Cloud resources it uses, which includes:

  • The Google Cloud resources such as compute instances, disks, networks, load balancers from Cloud Load Balancing, and Filestore instances. For example, in the context of a compute instance, the agent collects information such as the instance name and the region where it's deployed.
  • Information about the connections between the applications and databases in your SAP system.
  • Optionally, the names and versions of the SAP products used in your SAP system.

The collected information is provided to mandatory agent features such as SAP Host Agent metrics collection, other agent features that you enable, and Google Cloud services that you choose to use, such as Workload Manager evaluation. Consequently, this feature is always enabled and can't be turned off.

The following are the configurations that are supported for this feature:

  • The enable_workload_discovery boolean parameter determines whether or not the agent collects the names and versions of the SAP products running in your system. By default, this parameter is set to true.

  • The enable_discovery boolean parameter determines whether or not the collected information is sent to Cloud Logging. It's also sent to Workload Manager if you've enabled the Workload Manager API in your Google Cloud project. By default, this parameter is set to true.

    The information sent to Workload Manager lets you set up observability for your SAP workload. You don't incur any cost when the agent sends the collected information to Workload Manager. However, there are costs associated with the sending of data to Cloud Monitoring and Cloud Logging, which is required for enabling observability. For information about Workload Manager's observability service for SAP, including pricing, see SAP observability overview.

Agent health metrics collection

On Linux, Google Cloud's Agent for SAP can collect metrics that indicate the health of its daemon process on your host. This is an optional feature that you can enable after installing the agent.

This feature is not supported on Windows.

For instructions to enable the collection of the health metrics, and configure the collection frequency, see Configure agent health metrics collection.

The following table describes the health metrics that the agent collects. The metric strings in this table must be prefixed with workload.googleapis.com/. This prefix has been omitted from the entries in the following table.

Metric Category Description
sap/agent/health Agent for SAP metrics Indicates if the agent is actively running or not:
  • The value true indicates that the agent is active and running.
  • The value false indicates that the agent is not running.
sap/agent/cpu/utilization Agent for SAP metrics CPU utilization by the agent's daemon process.
sap/agent/memory/utilization Agent for SAP metrics Memory utilization by the agent's daemon process.

Process Monitoring metrics collection

On Linux, Google Cloud's Agent for SAP can collect Process Monitoring metrics for your SAP applications and their runtime states. This is an optional feature that you can enable after installing Google Cloud's Agent for SAP.

This feature is not supported on Windows.

The information collected in the Process Monitoring metrics helps you troubleshoot the issues related to your SAP system. In case of issues, with the help of Process Monitoring metrics, Cloud Customer Care can help you reach a resolution more efficiently. The data collected using Process Monitoring metrics provide observability for your SAP HANA high-availability cluster configurations.

Overall, Process Monitoring metrics are fast-changing metrics. For more information about this feature, see Process monitoring using Google Cloud's Agent for SAP.

Workload Manager evaluation metrics collection

On Linux, Google Cloud's Agent for SAP can collect Workload Manager evaluation metrics, which enable you to use Workload Manager to evaluate your SAP workloads against best practices.

From agent version 3.2, this feature is enabled by default for new installations.

This feature is not supported on Windows.

The information collected in the Workload Manager evaluation metrics help you check the configuration and settings of your SAP applications, databases, and high-availability configurations.

Workload Manager evaluation metrics are slow-changing metrics that are collected every 5 minutes by default. From agent version 3.2, you don't incur any additional cost for using the agent to collect the Workload Manager evaluation metrics.

For more information about this feature, see Workload Manager evaluation using Google Cloud's Agent for SAP.

SAP HANA monitoring metrics collection

On Linux, Google Cloud's Agent for SAP can collect metrics from your SAP HANA instances and send them to Cloud Monitoring. This optional feature of Google Cloud's Agent for SAP is the successor to using Google Cloud's monitoring agent for SAP HANA version 2.

This feature is not supported on Windows.

The SAP HANA monitoring metrics provide you visibility into the performance, availability, and health of your SAP HANA instances and the underlying infrastructure. The following are examples on how you can use these metrics:

  • You can use the memory utilization related metrics for capacity planning, or to help manage memory-based SAP HANA licensing.
  • By co-relating the SAP HANA monitoring metrics with the Compute Engine metrics, you can identify usage and performance trends.
  • By defining custom queries that collect custom metrics, you can gain deeper insights about your SAP HANA instances.
  • In Cloud Monitoring, you can create dashboards to visualize the SAP HANA monitoring metrics and set up alerts based on metric thresholds.

SAP HANA monitoring metrics are slow-changing metrics that, once enabled, are collected every 5 minutes by default.

For more information about this feature, see Monitoring SAP HANA using Google Cloud's Agent for SAP.

Backint based backup and recovery for SAP HANA

This feature of Google Cloud's Agent for SAP lets you back up and recover SAP HANA systems running on Google Cloud, on Bare Metal Solution, on premises, or on other cloud providers.

This feature of the agent is integrated with SAP HANA so that you can store and retrieve backups directly from Cloud Storage by using SAP-native backup and recovery functions.

For more information about this feature, see Backup and recovery for SAP HANA by using Backint.

Disk snapshot based backup and recovery for SAP HANA

This feature of Google Cloud's Agent for SAP lets you back up and recover SAP HANA systems running on Compute Engine instances by using disk snapshots.

Preview: To use this feature with SAP HANA systems that use striped disks to host the /hana/data volume, contact Customer Care. For information about how to contact Customer Care, see Getting support for SAP on Google Cloud.

For more information about this feature, see Backup and recovery for SAP HANA by using disk snapshots.

Workload performance diagnostics

To help Customer Care troubleshoot performance issues with your SAP workloads that run on Google Cloud, Google Cloud's Agent for SAP contains a built-in performance diagnostics tool. This feature is available from version 3.4 of the agent.

You can use this tool to run the following diagnostics. On-demand diagnostics need to be invoked, whereas default ones such as the guest OS diagnostics are run at all times. You can also have the tool run all these diagnostics in a single execution.

Diagnostic type Execution mode Description
Input/output diagnostics On-demand This is done by using the Flexible I/O tester (FIO) tool, which can stress-test the I/O capabilities of storage systems, and also simulate workloads to assess performance of the underlying infrastructure such as disks and network file system. For more information about FIO, see Flexible I/O tester (FIO).
Backup diagnostics On-demand This tests the performance of the Backint based backup and recovery operations for SAP HANA workloads. It includes testing your network connection and access to the Cloud Storage bucket that you've configured for use with the agent's Backint feature. For more information, see self diagnostic. It also runs the gsutil perfdiag diagnostic.
Guest OS diagnostics Default For workloads running on Compute Engine bare metal instances, the tool checks whether the guest OS is configured for the optimal running of SAP workloads. This is done by running the configureinstance -check command. For more information, see Configure guest OS on bare metal instances.

The output of this tool is a ZIP file that contains the diagnostic information that you specify the tool to gather.

For information about how to use this tool to gather diagnostic information for sharing with Customer Care, see Gather diagnostic information about SAP workload performance.

Pricing

Google Cloud's Agent for SAP is free to install and run. However, you are responsible for the charges that result from reading and writing metric data to Cloud Monitoring.

For pricing information related to the Backint feature of the agent, see Monthly cost estimate.

For the metrics collected by the agent, Monitoring charges according to the following:

  • Monitoring API read calls that exceed the free monthly API allotment.
  • Metrics measured by bytes ingested, when the ingested metric data exceeds the free monthly metric allotment.

In Monitoring, ingestion refers to the process of writing time series to Monitoring. Each time series includes some number of data points; those data points are the basis for ingestion charges. For the pricing information, see Monitoring Pricing.

For information about the general concepts in Monitoring, see Metrics, time series, and resources.

Monthly cost estimates

This section illustrates how to estimate monthly costs for collecting metric data for metrics charged by API read calls and bytes ingested. For estimating the cost of using the Backint feature of Google Cloud's Agent for SAP to perform backup and recovery operations for SAP HANA, see Monthly cost estimate.

The examples in this section are based on Monitoring pricing as of December, 2022. These examples are intended to illustrate calculations; for comprehensive estimates, use the Pricing Calculator.

The following table summarizes monthly cost estimates for some required and optional features of the agent. For the detailed steps for cost estimation, see SAP Host Agent metrics, Process Monitoring metrics, and SAP HANA monitoring metrics.

Feature Type Number of Compute Engine instances Cost ($)
SAP Host Agent metrics collection Required 100 $209
Process Monitoring metrics collection Optional 100 $564.81
SAP HANA monitoring metrics collection Optional 500 $280.30

Pricing example for metrics charged by API read calls

The following example illustrates how to get an estimate of costs for collecting metric data for metrics charged by API read calls, such as SAP Host Agent metrics.

SAP Host Agent metrics

Consider a scenario where you have some number of Compute Engine instances that make some number of API read calls each month. The variables in the scenario include the following:

  • The rate at which a compute instance makes API read calls.
  • The number of compute instances.

For this cost estimation, consider there are roughly 43,800 minutes in a month.

60 * 730 hours (365 days / 12 months * 24 hours)

Assume that a compute instance makes API read calls at the rate of 5 calls per minute.

For one compute instance, total API read calls at the rate of 5 calls per minute for one month: 219,000 (43,800 * 5)

Consider that you have 100 compute instances.

  • Total number of API read calls per month: 21,900,000

    (219,000 * 100)

  • Number of API read calls per month after free allotment: 20,900,000

    (21,900,000 - 1,000,000 = 20,900,000)

  • Approximate cost per month for 100 compute instances: $209

    (20,900,000 / 1000) * 0.01

No. of reads Rate ($/1000 reads) Cost ($)
Free allotment/month 1,000,000 0.00 $0.00
Chargeable 20,900,000 0.01 $209
Total 21,900,000 $209

Pricing example for metrics charged by bytes ingested

The following examples illustrate how to get a cost estimate for collecting metric data for the metrics charged by bytes ingested, such as Process Monitoring metrics and SAP HANA monitoring metrics.

Process Monitoring metrics

Consider a scenario where you have some number of Compute Engine instances that are writing data from some number of metrics each month. The variables in the scenario include the following:

  • The rate at which the metric data is written.
  • The number of metrics.
  • The number of Compute Engine instances.

For this cost estimation, consider that there are roughly 15 Process Monitoring metrics being collected, of which 11 are slow-changing metrics and 4 are fast-changing metrics. Also, consider that there are roughly 43,800 minutes in a month.

60 * 730 hours (365 days / 12 months * 24 hours)

For one fast-changing metric, writing data at the rate of 1 data point per 5 seconds, that is, 12 data points per minute for one month:

For one slow-changing metric, writing data at the rate of 1 data point per 30 seconds, that is, 2 data points per minute for one month:

  • Total fast-changing metrics data points is: 525,600 (43,800 minutes * 12 data points)
  • Total slow-changing metrics data points is: 87,600 (43,800 minutes * 2 data points)
  • Total volume ingested for each fast-changing metric is:
    • 4,204,800 bytes (525,600 data points * 8 bytes)
    • This is 4.01000977 MiB (4,204,800 bytes / 1,048,576 bytes/MiB)
  • Total volume ingested for each slow-changing metric is:
    • 7,00,800 bytes (87,600 data points * 8 bytes)
    • This is 0.66833496 MiB (7,00,800 / 1,048,576 bytes/MiB)

Consider that you have 100 compute instances, each writing 15 metrics.

  • Total monthly ingestion for fast-changing metrics: 1604.03908 MiB

    4.01000977 MiB for one fast-changing metric * 400 (100 compute instances * 4 metrics)

  • Total monthly ingestion for slow-changing metrics: 735.168456 MiB

    0.66833496 MiB for one slow-changing metric * 1100 (100 compute instances * 11 metrics)

  • Total monthly ingestion: 2339.207536 MiB

    Total monthly ingestion for fast-changing metrics + Total monthly ingestion for slow-changing metrics

  • Approximate cost per month for 100 compute instances: $564.81

    (2339.207536 - 150) * 0.258

MiB ingested Rate ($/MiB) Cost ($)
Free allotment/month 150 0.00 $0.00
Chargeable 2189.207536 0.258 $564.81
Total 2339.207536 $564.81

SAP HANA monitoring metrics

Consider a scenario where you have some number of Compute Engine instances that are writing data from querying SAP HANA databases each month. The variables in the scenario include the following:

  • The rate at which the metric data is written.
  • The number of queries.
  • The number of metrics collected for each query.
  • The number of Compute Engine instances deployed.
  • The number of SAP HANA database deployments per compute instance.

By default, Google Cloud's Agent for SAP has 12 built-in queries as part of its SAP HANA monitoring feature. The agent runs these queries once every 300 seconds (the default sample interval) on each SAP HANA database instance. These 12 queries generate a total of 37 metrics. Assume that each compute instance hosts only one SAP HANA database. Therefore, each compute instance writes a total of 37 metrics every 300 seconds.

For this cost estimation, consider that there are roughly 43,800 minutes in a month.

60 * 730 hours (365 days / 12 months * 24 hours)

For one metric, writing data at the rate of 1 data point per 300 seconds, that is, 0.2 data points per minute for one month:

  • Total data points is: 8,760 (43,800 minutes * 0.2 data points)
  • Total volume ingested is:
    • 70,080 bytes (8,760 data points * 8 bytes)
    • This is 0.0668335 MiB (70,080 bytes / 1,048,576 bytes/MiB)

Consider that you have 500 compute instances, each writing 37 metrics.

  • Total monthly ingestion: 1236.419678 MiB

    0.0668335 MiB for one metric * 18,500 (500 compute instances * 37 metrics)

  • Approximate cost per month 500 compute instances: $280.30

    (1236.419678 - 150) * 0.258

MiB ingested Rate ($/MiB) Cost ($)
Free allotment/month 150 0.00 $0.00
Chargeable 1,086.419678 0.258 $280.30
Total 1,236.419678 $280.30

Resources consumed by the agent

To help you evaluate the resources that Google Cloud's Agent for SAP consumes for collecting metrics, this section shows sample usages of the agent and the resources that they consumed in Google Cloud's test system with the following specifications:

  • Machine type: n2-highmem-64
  • Type of persistent storage disk attached to the Compute Engine instance: SSD Persistent Disk (pd-ssd)
  • SAP systems running on the compute instance: One instance of SAP HANA and two instances of SAP NetWeaver.

Note that the information in the following table is indicative and that the resources consumed by your agent instances can vary.

Example 1 Example 2 Example 3 Example 4
Compute instance runtime 24 hours 24 hours 24 hours 24 hours
Is SAP Host Agent metrics collection enabled? Yes Yes No* No*
Is Process Monitoring metrics collection enabled? Yes, at the default interval of 5 seconds Yes, at the default interval of 5 seconds No No
Is Workload Manager evaluation metrics collection enabled? Yes Yes Yes No
Is SAP HANA monitoring metrics collection enabled? Yes No No Yes, at the default interval of 300 seconds
CPU usage Less than 520 milliseconds per minute of a vCPU. This stays constant even as more vCPUs are added. Less than 500 milliseconds per minute of a vCPU. This stays constant even as more vCPUs are added. Less than 30 milliseconds per minute of a vCPU. This stays constant even as more vCPUs are added. Less than 20 milliseconds per minute of a vCPU core. This stays constant even as more vCPUs are added.
Memory usage Less than 40 MB Less than 50 MB Less than 40 MB Less than 29 MB
Disk writes Less than 70 KB per minute† Less than 70 KB per minute† Less than 6 KB per minute† Less than 5 KB per minute†
Disk reads Zero Zero Zero Zero
Network usage (steady state) Less than 2 MB per minute Less than 2 MB per minute Less than 600 bytes per minute Less than 1 MB per minute

* SAP Host Agent metrics collection is disabled only to test the consumption of resources. You must not disable the collection of SAP Host Agent metrics because it is mandated by SAP.

† Agent logs set at the INFO level.

Supported operating systems

Google Cloud's Agent for SAP supports all operating systems that are supported by SAP NetWeaver and SAP HANA. For details, see the list of supported operating systems for SAP NetWeaver and list of supported operating systems for SAP HANA.

Authentication and access

Google Cloud's Agent for SAP requires an Identity and Access Management (IAM) service account for authentication with Google Cloud and for permission to access Google Cloud resources.

IAM roles that are assigned to the service account determine which Google Cloud resources the agent has permission to interact with.

Unless you use an existing or default service account, you need to create the service account in your Google Cloud project and assign one or more roles to the service account.

Specify a service account

If the agent is running on a Compute Engine instance, then the agent uses the service account of the host compute instance by default. The host compute instance's service account is usually a service account with limited roles that you or one of your colleagues create, but it can instead be the Compute Engine default service account, which has broad permissions by default.

If the agent is not running on a Compute Engine instance, then you need to create a service account with a JSON service account key. For information, see Set up a service account.

For more information about service accounts, roles, and permissions, see:

Required IAM roles

Whether you use a new, existing, or default service account, the service account must include roles that grant the required permissions to the agent.

Feature Required IAM roles
Collection of SAP Host Agent metrics

For collecting SAP Host Agent metrics on Compute Engine instances, the following predefined IAM roles are required:

For collecting SAP Host Agent metrics on a Bare Metal Solution server, the agent does not access the Google Cloud APIs, and so it does not require an IAM service account or permissions.
Collection of Workload Manager evaluation metrics

The following predefined IAM roles are required:
Collection of Process Monitoring metrics or the SAP HANA monitoring metrics

The following predefined IAM roles are required:
Backint based backup and recovery for SAP HANA

Access to Google Cloud APIs

Compute Engine recommends configuring your instances to allow full access to all Cloud APIs and using only the IAM permissions of the instance service account to restrict access to Google Cloud resources. For more information, see Create a VM that uses a user-managed service account.

If you limit access to the Cloud APIs on a compute instance, then:

  • For the collection of SAP Host Agent metrics, Google Cloud's Agent for SAP requires the following minimum Google Cloud API access scopes on the host compute instance:
    • Compute Engine: Read Only
    • Stackdriver Monitoring API: Read Only
  • For the collection of Workload Manager evaluation metrics, or retrieval of credentials from Secret Manager, the Cloud Platform access scope must be enabled.
  • For collecting the Process Monitoring metrics or the SAP HANA monitoring metrics, the access scopes of the host compute instance must have the write access to publish metric data to your Google Cloud projects. When you create a new Compute Engine instance, it is automatically configured with the write access scope, https://www.googleapis.com/auth/monitoring.write.
  • For information about the Cloud APIs that you need to access for using the agent to perform backup and recovery operations for SAP HANA, see Enable access to Cloud APIs and metadata servers.

If the agent is not running on a Compute Engine instance, then you need to establish connection to Google Cloud APIs. See Configuring Private Google Access for on-premises hosts for details.

Metrics in Monitoring

For the Process Monitoring metrics and the SAP HANA monitoring metrics, Google Cloud's Agent for SAP sends the data that it retrieves from your SAP systems to Monitoring as custom metric data. To send the data, the agent uses the Monitoring API, which is enabled by default when you create a Google Cloud project.

For an overview of Monitoring, see Introduction to Monitoring.

Find the metrics data in Monitoring

The metrics data from Google Cloud's Agent for SAP is available to view as soon as Monitoring receives it. To find and visualize the collected data, you can use Metrics Explorer. For more information about finding the metrics data in Monitoring, see Google Cloud's Agent for SAP Operations guide.

For more information about visualizing the collected data in Monitoring, see:

Dashboard definitions on GitHub

On GitHub, you can find the definitions for custom dashboards that help you visualize the metrics collected by Google Cloud's Agent for SAP in Cloud Monitoring. For more information, see:

Alert notifications

To alert you when a metric reaches a threshold that you define, you can configure alert notifications in Monitoring.

For more information, see Introduction to alerting.

Automatic installation

When you use the Terraform configuration files that Google Cloud provides for automating the deployment of your Google Cloud infrastructure, Google Cloud's Agent for SAP is automatically installed as part of the deployment process. This is done to satisfy the SAP mandate of collecting the SAP Host Agent metrics.

For instructions for manually installing and configuring Google Cloud's Agent for SAP, see:

Google Cloud's Agent for SAP codebase on GitHub

For security requirements of your enterprise, if required, you can review the source code of Google Cloud's Agent for SAP. The codebase of the agent is available in a read-only GitHub repository, google-cloud-sap-agent under GoogleCloudPlatform/sapagent.

Support

For issues with Google Cloud infrastructure or services, contact Customer Care. You can find the contact information on the Support Overview page in the Google Cloud console. If Customer Care determines that a problem resides in your SAP systems, then you are referred to SAP Support.

For SAP product-related issues, log your support request with SAP support. SAP evaluates the support ticket and, if it appears to be a Google Cloud infrastructure issue, then SAP transfers that ticket to the appropriate Google Cloud component in its system: BC-OP-LNX-GOOGLE or BC-OP-NT-GOOGLE.

Support requirements

Before you can receive support for SAP systems and the Google Cloud infrastructure and services that they use, you must meet the minimum support plan requirements.

For more information about the minimum support requirements for SAP on Google Cloud, see:

Gather diagnostic information

Before contacting Customer Care for assistance in resolving issues related to Google Cloud's Agent for SAP, we recommend that you gather all relevant diagnostic information, notably the agent's support bundle.

For more information about gathering diagnostic information for the agent, see Google Cloud's Agent for SAP diagnostic information.

What's next