A placement group is a logical grouping of instances within a single availability zone. Placement groups enable applications to participate in a 10-Gbps network. Placement groups are recommended for instances that need low network latency, high network throughput, or both. Teradata recommends using the same instance types within a placement group. Mixing instance types may result in AWS insufficient capacity errors.
Create placement groups before launching a Teradata Database instance, especially a Teradata Database MPP instance. If you do not use placement groups, database performance will be impacted. For example, you may experience high CPU skew on a node. Also use placement groups for Teradata Data Mover, Teradata Server Management, and Teradata DSC instances to ensure better performance.
After creating a placement group, you can launch multiple instances into the same group. For example, create a placement group containing the Teradata Database and Teradata DSC instances. After launching an instance, you should not modify the placement group.
When launching a Teradata ecosystem, you can use an existing placement group for Teradata Database, Teradata Data Mover, and Teradata Data Stream Controller. Use placement groups for these products, especially if you plan to move large amounts of data.
For more information on creating or deleting placement groups, or about placement group limitations, see Amazon User Guide for Linux Instances.