Comparison of migration tools

Use the table and the hints to decide which migration tool is more suitable for you.

Consider the following main differences:

The CDSW Migration tool approach requires a cutover of all CDSW users to Cloudera AI at once. In case of a large number of users and projects, this approach might pose a huge risk. In case of unexpected critical issues, rollback to CDSW is possible. However, the migration must be restarted from the beginning using a new workbench, since we cannot trigger a migration for the migrated workbench once it has completed the final migration.
The Cloudera AI command line utility tool allows migration of users and their projects in batches. This allows you to perform the migration at your own pace. However, more effort might be required to coordinate with the users.

Table 1. Comparison of migration tools
	Cloudera AI command-line utility	CDSW Migration tool
Migration approach	Project level	Cluster level
Migration effort	Progressive High effort required Anything outside the projects need to be handled manually.	Quick, complete Low effort required Migration is fully automated.
Downtime	No	Yes During final migration
What is migrated
Site-level settings (security, environment variables, quotas and so on)
Users		^*
Runtime images
Projects (models, jobs, applications, environment variables)
^* The CDSW migration tool replicates the CDSW PostgreSQL Database to the Cloudera AI Workbench database, which includes the user accounts. However, you will still need to create the users or groups on the Cloudera Management Console on premises (that is, User Management Service) to grant access to the migrated Cloudera AI Workbench.