Administration Guide › Scalability Best Practices › Scalability Recommendations › Remote Deployment and Policy Configuration Recommendations › Scalability Use Cases › Multiple Data Centers

Multiple Data Centers

In this use case, 10,000 systems, spread across multiple Data Centers, are being managed. The number of systems per data center varies from 500 through 2,500. The Data Centers are geographically distributed across multiple locations. The links between the Data Centers include leased lines of less than 100Mb/sec. The systems run various workloads, and are managed by a number of different departments (application owners).

Recommendations for this environment are:

• Component Installation

  Install the CA Server Automation manager components on a dedicated server. This server must meet the minimum supported specification, but a quad-core server with at least 8-GB RAM is recommended.

  We recommend installing the Database on a separate dedicated server with an increased specification of quad-core processor and 8-GB RAM.

  To support the remote data centers, we recommend installing one Distribution Server in each data center. For a data center that contains more than 2,000 servers, we recommend installing a second Distribution Server.

  Note: If you use Policy Configuration and you do not use Remote Deployment, each Distribution Server can scale up to 3,000 servers.

• Initial Deployment

  The following are relevant factors to consider while planning to deploy using multiple distribution servers:

  • If possible, limit a single deployment job size to a maximum of 500 target systems.
  • Verify that the nearest distribution server is chosen for the deployment operation.
  • Although concurrent deployments are supported within a single distribution server, it is advised to only use concurrent deployments across multiple distribution servers. If you have 4 distribution servers, you could start a job to deploy to 500 machines for every distribution server.
  • When you perform concurrent deployment using the same distribution server, verify that the second job does not deploy to the same set of target machines as the previous one.
  • Where multiple packages are being delivered to many systems, consider splitting the package deliver to multiple jobs. For example, by deploying SystemEDGE and Advanced Encryption first.
  • If you do see failures during a large-scale deployment, verify all prerequisites then use “resubmit job” to retry the deployment.
  • To help with future deployments, we recommend saving the deployments as templates.
• Service membership

  For ease of maintenance, we suggest aligning the monitored servers into Services, with a maximum of 500 servers per Service. For remote data centers, we recommend creating one or more Services (depending on data center size) to represent data center.

  Where multiple departments use particular systems, the systems can be added to multiple services for ease of management by each department.

• Applying Policies

  In this use case, the servers run various workloads. We therefore recommend that the base policy contains only the control settings. Hold your monitoring configuration in templates based on different monitoring requirements. These templates can then be applied to the required systems, either by manually selecting systems, or by applying templates to a service.

  Splitting the monitoring requirements into templates allows the templates to be applied to the required systems, independently of each other. Select the systems manually, or apply templates to a service. We recommend applying templates in batches of 2,000 to 2,500 systems.

  If the base policy is required to be changed, we recommend applying the policy to systems in batches of 2,000 to 2,500 systems.

  Note: When templates are used, each delivery of a template or policy involves merging of all assigned templates with the base policy. The next step is delivering the resultant configuration to the agent. Therefore, where multiple templates are applied to a system, the time for delivery may be slightly increased.