SRDF is PowerMax remote replication software that offers extensive flexibility, scale, and features, while maintaining high performance and low latencies.
An SRDF group is a collection of SRDF ports from two PowerMax systems connected by an FC or IP-based network. When source and target volumes are added to the SRDF group (for example, using SGs for ease of management), a replication session can be started to obtain a copy of the data between the PowerMax systems. PowerMax systems support up to 2,048 SRDF groups (up to 250 groups if one of the systems is an older generation PowerMax system).
SRDF supports many topologies:
- A topology with two storage systems has flexibility of operations such as disaster protection across synchronous or asynchronous distances, data mobility, and migration.
- A topology with three storage systems has the use of cascaded or STAR-type replications.
- A topology with four storage systems can be used for migrations.
The following key SRDF modes include:
- SRDF/Synchronous (SRDF/S)—This mode ensures that any write I/O to the source volumes is replicated to the SRDF target volume immediately before the I/O is acknowledged on the replication source of the host. This mode allows for a disaster restart solution with zero data loss of committed transactions. If the replication source becomes unavailable, the application can restart from the target volumes, rolling back (removing) uncommitted database transactions, and rolling forward (reinstating) any outstanding committed transactions automatically. Synchronous replications affect host writes latency based on the distance, I/O size, network speed, and network quality. However, because PowerMax cache is considered persistent, the remote PowerMax system acknowledges the writes when they register with its cache.
- SRDF/Asynchronous (SRDF/A)—This mode also provides a disaster restart solution, as the target volumes maintain data consistency. However, I/Os to the replication source volumes are acknowledged immediately to the host. Data changes (writes) are aggregated into consistent cycles (batches), which are sent over the SRDF links to the remote storage system. The remote storage system can only enable full (consistent) cycles to be written to the volumes, ensuring the remote data is always consistent, though slightly lagging behind the source. SRDF/A provides a remote replication solution that maintains data consistency without performance overhead at any distance. The remote data will be slightly behind the source due to the nature of cycle shipping.
- SRDF/Metro—This mode is a variation of SRDF/S mode. Instead of having a source and target for the remote replication, when the system is made consistent, both storage systems can replicate writes in both directions. SRDF/Metro allows creation of stretch clusters and increased availability across distances. If there is a disaster, SRDF/Metro uses a virtual Witness (vWitness) to determine in real time which system remains active. With PowerMaxOS 10, the vWitness is Solutions Enabler vWitness software running in a Linux host environment. SRDF/Metro can be configured to add a single SRDF/A target from each of the storage systems with a solution called SRDF/Metro SmartDR.
- SRDF/Adaptive-Copy (SRDF/ACP)—This mode is a data mobility mode that does not maintain data consistency. It can be used for bulk transfer of data that is already consistent (such as a consistent snapshot or clone). This mode can also match an SRDF target with the source volumes’ data in small batches until the data is close to the source volumes. At that point, the replication mode can be changed to SRDF/S or SRDF/A.
For more information about SRDF/Metro, see the Dell PowerMax and VMAX All Flash: SRDF/Metro Overview and Best Practices White Paper.