Cabling for replication
Connecting two storage systems to replicate
volumes
The replication feature performs asynchronous replication of block-level data from a volume in a primary system to a volume in a
secondary system.
Replication creates an internal snapshot of the primary volume, and copies the changes to the data since the last replication to the
secondary system using FC or iSCSI links.
The two associated standard volumes form a replication set, and only the primary volume (source of data) can be mapped for access by a
server. Both systems must be connected through switches to the same fabric or network (no direct attach). The server accessing the
replication set is connected to the primary system. If the primary system goes offline, a connected server can access the replicated data
from the secondary system.
Systems can be cabled to support replication using CNC-based and 10Gbase-T systems on the same network, or on different networks.
NOTE:
SAS systems do not support replication.
As you consider the physical connections of your system, keep several important points in mind:
•
Ensure that controllers have connectivity between systems, whether the destination system is colocated or remotely located.
•
Qualified Converged Network Controller options can be used for host I/O or replication, or both.
•
The storage system does not provide for specific assignment of ports for replication. However, this configuration can be accomplished
using virtual LANs for iSCSI and zones for FC, or by using physically separate infrastructure.
•
For remote replication, ensure that all ports that are assigned for replication can communicate with the replication system by using the
query peer-connection CLI command. See the
ME4 Series Storage System CLI Reference Guide
for more information.
•
Allow enough ports for replication permits so that the system can balance the load across those ports as I/O demands rise and fall. If
controller A owns some of the volumes that are replicated and controller B owns other volumes that are replicated, then enable at
least one port for replication on each controller module. You may need to enable more than one port per controller module depending
on replication traffic load.
•
For the sake of system security, do not unnecessarily expose the controller module network port to an external network connection.
Conceptual cabling examples are provided addressing cabling on the same network and cabling relative to different networks.
NOTE:
The controller module firmware must be compatible on all systems that are used for replication.
Host ports and replication
ME4 Series Storage System controller modules can use qualified 10Gbase-T connectors or CNC-based ports for replication.
CNC ports must use qualified SFP+ transceivers of the same type, or they can use a combination of qualified SFP+ transceivers
supporting different interface protocols. To use a combination of different protocols, configure host ports 0 and 1 to use FC, and configure
ports 2 and 3 to use iSCSI. FC and iSCSI ports can be used to perform host I/O or replication, or both.
NOTE:
ME4 Series 5U84 enclosures support dual-controller configurations only. ME4 Series 2U controller enclosures
support single controller and dual-controller configurations.
• If a partner controller module fails, the storage system fails over and runs on a single controller module until the
redundancy is restored.
• In dual-controller module configurations, a controller module must be installed in each slot to ensure sufficient
airflow through the enclosure during operation. In single-controller module configurations, a controller module must
be installed in slot A, and controller module blank must be installed in slot B.
A
76
Cabling for replication
