Creating a Mirrored Volume in a Remote Storage with REC (Asynchronous Consistency Mode)

This section describes operations that use the REC feature in asynchronous consistency mode to create a mirrored volume of operation volumes (volume with the same data as the operation volume) in a remote storage system. When an REC Session in asynchronous consistency mode writes data to a operation volume, it asynchronously transfers data to the mirrored volume after responding that the data write is complete. By performing periodic data transfers so that the order when writing data to the operation volume and the order when transferring data to the mirror volume are the same, the data in the mirrored volume is identical to the data in the operation volume a few seconds to few minutes before the operation. In addition, since the data transfer is asynchronous with the data writing to the operation volume, the degradation of the response performance to the operation volume can be prevented.

Caution

  • To create an REC Session in asynchronous consistency mode, the REC buffer (dedicated buffer for order assurance) must be configured. In addition, if the amount of data updated to the operation volume exceeds the line bandwidth of the copy path, the REC buffer may be exhausted. To avoid running out of REC buffers, REC disk buffers are available.

  • If the line bandwidth of the copy path is low or the operation volume updates is large, consider using an asynchronous stack mode REC Session for backup operations.

  • Because data is transferred asynchronously to the mirrored volumes, the Recovery Point Objective (RPO) can vary from few seconds to few minutes. If you want a zero RPO, consider using a synchronous REC Session for backup operations.

RESTful API and Sample Scripts for Operation Use

  • Preparation

    Create a volume on the remote storage system that is the same size as the operation volume, and create an REC Session (Asynchronous Consistency Mode) between the operation volume and the created volume. Use rec_consistency_create.py.

  • Operation

    No action is required. After the initial copy of the REC Session is complete, updates written to the operation volume are periodically transferred to the mirrored volume with guaranteed ordering.

    If you want to temporarily switch the transfer mode to Stack to suppress the REC buffer Halt, you can use PATCH /copysession/{copysession_id} API.

  • Restore

    When restoring operation volume data from a mirrored volume, the REC Session is reversed and data is transferred from the remote to the local storage system, after recovery of the copy path and RAID Group if necessary. For REC Session inversion, rec_sync_consis_recover.py can be used. Before inversion, the local storage system must be configured with a receive REC buffer and the remote storage system with a transmit REC buffer.

  • Deletion

    If you want to remove a mirrored volume, force delete the REC Session, and then remove the destination volume. For forced deletion of the Copy Session, use DELETE /copysession/{copysession_id} API. For deletion of volumes, use DELETE /volume API. For more information, see the description of Deleting a Copy Session or Deleting Multiple Volumes.

    To remove only the REC Session and continue to use the mirrored volume in the remote storage system, remove only the REC Session. An REC Session in asynchronous consistency mode cannot be deleted by using DELETE /copysession/{copysesion_id} API. Use copysession_delete.py.

Notes

  • If the operation volume is frequently updated, the response performance of the operation volume may be degraded due to the priority of data transfer processing.

    If the REC buffer becomes too full to hold new forwarding data due to excessive operation volume updates or due to copy path instability, the data transfer process is interrupted so that the response to the server is not delayed beyond a certain time. In this case, the REC Session will temporarily transition to Halt (REC buffer Halt). REC Buffer Halt automatically resumes data transfer after a few minutes if Recovery Mode is set to Automatic, but data written during REC Buffer Halt cannot be guaranteed order. If the response performance deteriorates or REC buffer Halt occurs chronically, the environment should be reviewed.

  • Consider introducing REC disk buffers to prevent REC buffer exhaustion due to temporary increases in operation volume updates and temporary instability in copy paths.

    For more information on how to do this, see "Assign REC Disk Buffer" in "Web GUI User's Guide".