EMC – SRDF Theory

Print Friendly

EMC SRDF (Symmetrix Remote Data Facility) is a replication product which can be used to replicate the data from one array to second array. Primary use is for business continuity/disaster recovery. Other use is migrating the data from one array to another. This post contains brief explanation of terms associated with SRDF, its operational use, and the commands commonly used.

RA port and Group:

RA Port – used for replication between EMC arrays
RA Group – a group of RA ports between 2 EMC arrayss. Each RA group has a unique group id.

Various Link status:

RW – Ready – Enabled for both reads and writes
NR – Not Ready – Disabled for both reads and writes
WD – Write Disabled – Enabled for reads but not writes
NA – Not available – Unable to report on correct state

Different modes of SRDF

DISCLAIMER: Below information has been compiled from different pdfs. Please read them just to get an overview of different types of SRDF replication.

Synchronous replication (SRDF/S):

In this mode of the operation, R2 is a real time mirror image of R1. Data on R1 and R2 are always fully synchronized. The write I/O is received from local host into cache of source array. It is then transferred to the cache of the remote array. Upon successful receipt of write in its cache, remote array sends an acknowledgement to source array. Source array in turn acknowledges successful ending status to local host.

Asynchronous replication (SRDF/A):

In this mode, host writes are accumulated on source array, and host receives acknowlegement for all writes. When the cycle time is reached, all accumulated data is transferred to remote array in one delta set. It uses 4 types of Delta Sets to manage the IOs. These are as below:

Capture Delta set on Source Array – It captures all incoming writes to all R1s defined in SRDF/A group.
Transmit Delta Set on Source Array – It transfers its contents from the source to the remote array.
Receive Delta Set on Remote Array – It receives the data being transferred by the source-side Transmit Delta Set.
Apply Delta Set on Remote Array – It writes the delta sets to R2s defined in SRDF/A group to create a consistent recoverable remote copy.

Above cycle is repeated to provide a continuous checkpoint of delta sets. Data on remote array is within the range of seconds to minutes of source array. Delta sets are resident in cache and hence requires extra additional cache. Approximately 0.75GB additional cache per 1TB of data being remotely mirrored is the minimum requirement. Extra space for delta set space can be assigned from hard drives using the feature called as “Delta Set Extension”.

Since the writes are transferred in cycles, any duplicated written to can be eliminted through Symmetrix ordered write processing, which transfers the changed tracks over the links only once within any single cycle.

Enginuity 5874 or higher incorporates a feature called as “Write Pacing”. When the SRDF I/O service rates are lower than the host I/O rates, it takes corrective action to slow host I/O rates to match the SRDF I/O service rates. It is a dynamic feature applied at the SRDF group level. This helps control the amount of cache on source array used by SRDF/by preventing it from getting exhausted.

Adaptive Copy replication (SRDF Data Mobility – SRDF/DM):

In this mode, the source R1s and target R2s are a few or many IOs out of synchronization. The write I/O is received from local host into cache of source array. Source array acknowledges successful ending status to local host. Source array processes the IO using either Adaptive Copy Write Pending Mode or Adaptive Copy Disk mode. It is then placed into SRDF queue, and transmitted to cache of the remote array. Upon successful receipt of write in its cache, remote array sends an acknowledgement to source array.

There are two ways in which Adaptive Copy can operate – Adaptive Copy Disk Mode and Adaptive Copy Write Pending Mode.

Adaptive Copy Disk Mode

Source array acknowleges all the writes to hosts, and saves them in cache of local array. These writes are de-staged from cache to source volume (R1s) and these IOs are tracked as invalid tracks. The data is subsequently transferred to remote volume (R2s).

Adaptive Copy Write Pending Mode

Source array acknowledges all the writes to hosts. These writes are saved in the cache of local array until it is successfully written to both the source volumes (R1s) and the target volumes (R2s).

SRDF/DM is a data replication solution (data migrations and data center moves) instead of disaster recovery solution.

SRDF/Automated Replication (SRDF/AR):

It is an automated replication solution that uses both SRDF and TimeFinder to provide a periodic asynchronous replication of a re-startable data image. It is offered in two varieties – SRDF/AR Single Hop and SRDF/AR Multiple Hop.

SRDF/AR (Single Hop)

In this scenario, BCVs of source devices at local site act as R1s and they replicate to R2 devices at remote site which might have its own BCV devices.

SRDF/AR (Multiple Hop)

3 sites are involved in this operation. Source devices from primary site are replicated to R2 devices at intermediate site via SRDF/S. BCV devices of these R2 devices acts as R1s and they replicate to R2 devices at target site using SRDF/AR Single Hop.

SRDF/Star Using Concurrent SRDF Mode:

3 sites are involved in this operation. Primary site has R11 devices which are source devices for remaining 2 sites, secondary site has R21 devices, and tertiary site has R22 devices. R22 volumes have 2 SRDF mirrors, only one of which is allowed to be active at a given time. Using R22 involves below mentioned steps:

1. Create R1/R2 pair between primary and secondary sites using SRDF/S
2. Create R11/R2 pair between primary and tertiary sites using SRDF/A
3. Create R21/R22 pair between secondary and tertiary sites using SRDF/A.

In case of primary site failure – failover to secondary site making it primary, and resume remote mirroring to tertiary site.

SRDF/Extended Distance Protection (SRDF/EDP):

It involves 3 sites and provides ‘No Data Loss’ extended distance disaster protection solution. It uses the basic cascaded SRDF configuration. Device R21 is a cache only device and there is no disk copy of data. R1 at primary site and R3 at tertiary site are only two full copies of data.

, , , , ,

No comments yet.

Leave a Reply