Data Lifecycle Manager terminology
Data Lifecycle Manager (DLM) is a UI service that is enabled through DPS Platform. From the DLM UI you can create and manage replication and disaster recovery policies and jobs.
| Term | Description |
|---|---|
| DLM App or Service | The web UI that runs on the DPS platform host. The corresponding agent needs to be installed on the clusters. |
| DLM Engine | The agent required for DLM. Also referred to as the DLM Engine, this replication engine must be installed as a management pack on each cluster to be used in data replication jobs. The engine maintains, in a configured database, information about clusters and policies that are involved in replication. |
| Data center | The facility that contains the computer, server, and storage systems and associated infrastructure, such as routers, switches, and so forth. Corporate data is stored, managed, and distributed from the data center. In an on-premise environment, a data center is often composed of a single HDP cluster. However, a single data center can contain multiple HDP clusters. |
| IaaS cluster | A full HDP cluster on cloud VMs with Apache services running, such as HDFS, YARN, Ambari, Hiveserver2, Ranger, Atlas, and DLM Engine. Replication behavior is similar to on-premise cluster replication. The data is on local HDFS. |
| Cloud data lake or data lake | An HDP cluster on the cloud, using VMs, with data retained on cloud storage. A cloud data lake requires minimal services for metadata and governance, such as Hive metastore, Ranger, Atlas, and DLM Engine. |
| Cloud storage | Any storage retained in a cloud account, such as Amazon S3 web service. |
| On-premise cluster | A full HDP cluster in a data center, with Apache services running, such as HDFS, Yarn, HMS, hiveserver2, Ranger, Atlas and DLM Engine. Replication behavior is similar to IaaS cluster replication. The data is on local HDFS. |
| Policy | A set of rules applied to a replication relationship. The rules include which clusters serve as source and destination, the type of data to replicate, the schedule for replicating data, and so on. |
| Job | An instance of a policy that is running or has run. |
| Source cluster | The cluster that contains the source data that will be replicated to a destination cluster. Source data could be an HDFS dataset or a Hive database. |
| Destination cluster | The cluster to which an HDFS dataset or Hive database is replicated. |
| Target | The path on the destination cluster to which the HDFS dataset or Hive database is replicated. |