Open Replicator Notes Print
Written by Administrator   
Tuesday, 18 October 2011 08:39

Open Replicator Notes

Open Replicator Notes

Introduction

EMC Open Replicator enables remote point-in-time copies to be used for data mobility, remote vaulting, and migration between EMC Symmetrix VMAX (or Symmetrix DMX) and qualified storage arrays with full or incremental copy capabilities. Open Replicator can:

  • Pull from source volumes on qualified remote arrays to a Symmetrix VMAX (or Symmetrix DMX) volume.
  • Push any life source VMAX volume to a target volume on a qualified array with incremental updates.
  • Perform online data migrations from qualified storage to VMAX with minimal disruption to host applications

Definitions

Terminology: Control and remote

Open Replicator can transfer data between Symmetrix arrays (homogeneous) or between a Symmetrix VMAX (or Symmetrix DMX) and another qualified Fibre Channel array (heterogeneous). The Symmetrix VMAX (or Symmetrix DMX), where Open Replicator runs, and its devices are always referred to as the control side of the copy operation. Other Symmetrix arrays, CLARiiON arrays, or third-party arrays on the SAN are referred to as the remote array/device.

Control FA "host" setup requirements

Open Replicator runs as an application in the Fibre Director (FA) of the Symmetrix VMAX. The Open Replicator software causes the FA to appear as an open systems host to the remote storage array. Therefore, no special software is needed in the remote storage array. However, zoning and LUN masking is needed to access the remote devices and must be defined (as is the case anytime a host needs to access devices on a remote storage array). Configuring the required zoning and LUN masking is the most common area where problems occur when setting up Open Replicator and will be covered thoroughly in multiple examples. Additionally, multiple methods for verifying that the setup is correct will be covered in the examples.

Terminology: Push and pull, cold and hot

Open Replicator supports two types of copy operations: push and pull.

  • push operation copies data from the control device to the remote device.
  • pull operation copies data to the control device from the remote devices.

Open Replicator has two modes of operation:

  • cold (offline) and
  • hot (online).

These two modes refer to the state of the VMAX resident devices (control devices). The terms online or offline are used to indicate the potential state of a host application that uses these devices. Only with a hot operation can an application using VMAX based storage remain online. For data consistency reasons, the remote devices must never be written to by any host connected to the remote array. In cases where the remote device is the source of the Open Replicator copy operation (a pull operation), it may be permissible for a remote host to have read-only access to the remote device.

Hot Push

Open Replicator can push data volumes out from a Symmetrix either in a live mode (hot) or from a static copy or source volume (cold). For a live push, no local point-in-time copies of the volumes are required. The Symmetrix creates logical point-in-time copies without having to allocate additional disk space, and I/O is permitted against the source volume during the transfer. If the application attempts a write to a location whose original point-in-time data has not yet been copied to the remote device, then Copy on First Write (COFW) applies, delaying the host I/O until the data is safely on the remote. All FA ports for the control devices must be configured (zoned and LUN masked) to see the remote devices, so that the FA which encounters the not-yet-copied data can perform the COFW itself. The initial hot push must be a full copy, and often includes the option to save differential information. The saved differential information is used when the hot push is recreated and activated, pushing only incremental changes since the previous activate. For data migration, this type of full hot push followed by repeated incremental pushes would be customary, with the final incremental push occurring while the application is shut down, allowing the remote copy to be fully up-to-date.

Multipathing with hot setup requirements

Typically a Symmetrix device is visible to a host on more than one I/O path. This is done to improve both fault tolerance (failover) and performance (load balancing). Multipathing is accomplished by configuring the Symmetrix to present Symmetrix Logical Volumes (SLVs) as being visible to the application host on more than one FA port. The application host must use some type of multipathed I/O solution in order to correctly handle the same device being presented on more than one I/O path. A common multipathing host solution is EMC PowerPath®. The combination of multipathed control devices and hot Open Replicator operations result in the greater requirement that all FA directors that present the control devices, must be zoned and LUN masked as hosts able to access the remote array devices.