The site admin must be able to configure and deploy cluster storage for the NDS Labs Workbench. We will deploy multiple different clusters for development and testing purposes and will eventually deploy clusters at other sites (e.g., SDSC, TACC). It is assumed that the deployed cluster storage is distributed, reliable, and performant.

Actor: Site admin

Prerequisites:

• Site admin has an understanding of the size, reliability, and performance of the storage being provisioned
• Site admin has access to unallocated storage infrastructure required to support the allocation
• Site admin has access to allocate storage-specific cluster compute resources to host the storage
• Site admin can specify the storage system declaratively - detailing the final layout of cluster-volumes, nodes, infrastructure-volumes

Primary flow:

• Declare the layout and resource mappings
• Checkpoint the infrastructure
• Execute the layout - Provision resources, intitialize volumes and servers, apply and initialize volume services
• Unwind any partial allocation on error
• Bring volume services online - Start Gluster server processes
• Present the access paths and credentials to admin

Alternative flows:

• Expand cluster-wide storage (add new blocks, servers)
• resize an existing storage layout

Artifacts:

• Documentation describing the storage cluster configuration process
• An execution engine committed to a github repo - takes declaration and performs primary and alternate flows
• - ansible/other?