Manila System Architecture

The Shared File Systems service is intended to be ran on one or more nodes.

Manila uses a sql-based central database that is shared by all manila services in the system. The amount and depth of the data fits into a sql database quite well. For small deployments this seems like an optimal solution. For larger deployments, and especially if security is a concern, manila will be moving towards multiple data stores with some kind of aggregation system.


Below you will a brief explanation of the different components.

                                                    /- ( LDAP )
                                [ Auth Manager ] ---
                                       |            \- ( DB )

[ Web Dashboard ]- manilaclient -[ manila-api ] -- < AMQP > -- [ manila-scheduler ] -- [ manila-share ] -- ( shared filesystem )
                                    < REST >
  • DB: sql database for data storage. Used by all components (LINKS NOT SHOWN)

  • Web Dashboard: external component that talks to the api, implemented as a plugin to the OpenStack Dashboard (Horizon) project, source is here.

  • manila-api

  • Auth Manager: component responsible for users/projects/and roles. Can backend to DB or LDAP. This is not a separate binary, but rather a python class that is used by most components in the system.

  • manila-scheduler

  • manila-share

Further Challenges

  • More efficient share/snapshot size calculation

  • Create a notion of “attached” shares with automation of mount operations

  • Allow admin-created share-servers and share-networks to be used by multiple tenants

  • Support creation of new subnets for share servers (to connect VLANs with VXLAN/GRE/etc)

  • Gateway mediated networking model with NFS-Ganesha

  • Add support for more backends