• Containerized RDS: “Split-brain” in computing Storage Separation Architecture

  • Containerized RDS: Computational storage separation or local storage?

  • To generate the Volume

  • Mount the database instance to the node to start the database

  • Obtain the Volume from the storage pool, mount it to the specified Node, and start the persistent MySQL application on the Node

  • MySQL has rebuild/failover/rescheduling capabilities

  • To generate the Volume

  • Mount the database instance to the node to start the database

  • Database instance because recreate/failover/rescheduling, by scheduling the new node

  • Volume Is unmounted from the original node

  • Mount the database to the new node

  • Before mounting a Volume, notify the Cloud Provider

  • After the Volume is unmounted, you need to notify the Cloud Provider

  • To generate the Volume

  • Attach to the node where the database instance resides

  • Mount the database instance to the node to start the database

  • Database instance because recreate/failover/rescheduling, by scheduling the other nodes

  • Volume Is unmounted from the original node

  • Volume detach from the original node

  • Attach to the new node

  • Mount the database to the new node

  • Before Volume is mounted, Kubelet will confirm whether it has been attached

  • Volume Before detach, the AttachDetach Controller confirms whether it has been unmounted

  1. Node restart/Crash;

  2. The network between the Node and API Server is faulty.

  3. Kubelet could not get CPU time fragments under high load.

  4. , etc.

    Click on the specific training content