MicroStrategy ONE
Architecture Best Practices and Details
General Architecture Details
- Only one Platform Analytics Consumer (Telemetry Store) can be installed per machine and per environment (when having one Platform Analytics Repository); hence multiple instances or versions of the Telemetry Store cannot be running in the same hosting machine.
- Multiple Telemetry Stores could potentially be running in different machines when each of those Telemetry Stores are pointing to different Platform Analytics repositories.
- The Telemetry Store consumer and the Intelligence Telemetry producer have no direct dependency on each other. The Telemetry Store communicates with the Telemetry Server/Telemetry Manager, Telemetry Cache, and the Platform Analytics Repository. The Intelligence Telemetry producer only sends data to the Telemetry Manager/Telemetry Server cluster.
- By default, when you install Platform Analytics on the Intelligence Server, the MicroStrategy installer will also install a single node of the Telemetry Server and Telemetry Manager.
- The Telemetry Store will prevent two consumer instances from running on the same machine in order to help reduce the risk of data loss and data integrity issues. However, customers should ensure that a separate Telemetry Store is not installed and running on a separate machine and configured to an already occupied Platform Analytics Repository.
- It is supported to have two Telemetry Store consumers (in different machines) writing to two separate PA Repositories as two independent Platform Analytics configurations.
- One Telemetry Store consumer is capable of processing the telemetry logs from two or more independent Intelligence Server clusters into one Platform Analytics Repository. This configuration provides a consolidated view of the statistics usage across multiple MicroStrategy environment. For configuration steps, see Monitor Metadata Repositories Across Multiple Environments.
- The Platform Analytics MicroStrategy project can be imported in any existing or newly created MicroStrategy metadata.
- For an estimation of resource requirements for stable and performant operation of the Platform Analytics components under a consistent transactional load, see KB482872: Capacity Planning for Platform Analytics.
Architecture Best Practices
- Platform Analytics supports installing all components on a single machine. However, this is not recommended for production environments where performance can be impacted by having limited machine resources being shared by all services.
- Have three clustered nodes as the minimum for the Telemetry Server/Telemetry Manager cluster for production environments or any environment where data redundancy and failure tolerance are critical. This will ensure that if a single Telemetry Server/Telemetry Manager node fails, the logs produced by the Intelligence Telemetry producer are persisted on the remaining two nodes.
- Have an odd number of clustered nodes of the Telemetry Server/Telemetry Manager. This ensures having a "majority rule" to prevent data loss in case of discrepancies and failures. For more information, see Apache Kafka Data Replication and ZooKeeper configurations for the Telemetry Server cluster.
- There must only be one Telemetry Store consumer writing data to one Platform Analytics Repository. If there are two Telemetry Store consumers writing data to the same PA Repository, there will be data loss and data integrity issues.
For examples of supported architectures, see Platform Analytics Architecture Examples.