Powermax Family Administration - Remote Replication: Participant Guide
Powermax Family Administration - Remote Replication: Participant Guide
ADMINISTRATION –
REMOTE REPLICATION
PARTICIPANT GUIDE
PARTICIPANT GUIDE
PowerMax Family Administration – Remote Replication-SSP New
SRDF Concepts 10
SRDF Overview 11
SRDF Use Cases 12
SRDF Terminology 13
SRDF Group Support Limits 14
SRDF Topologies 15
Two-Site Configurations 16
Concurrent SRDF Configuration 17
Cascaded SRDF Configuration 18
SRDF/Star Configuration 19
Four-Site Configuration 20
Knowledge Check 1 22
Knowledge Check 2 23
Appendix 143
SRDF Overview
SRDF Terminology
The commonly used SRDF terms and descriptions are shown in the table.
Term Description
R21 Devices that have a dual role and used in a cascaded three-
Devices site configuration with the following functionality:
• R21 device acts as an R2 device that receives updates
from the R1 device.
• R21 device acts as an R1 device that sends updates to
the R2 device.
Invalid Tracks which are not synchronized and not guaranteed with
Tracks consisitency between the two devices in an SRDF pair.
Unisphere SRDF Groups page showing SRDF group number 2048 (7FF).
Two-Site Configurations
Most the of the BFSI sectors running mission-critical applications use two-
site replication based on the distance between the sites and the
requirement of RPO and RTO for their organization.
Three-site replication is used when customers are looking for data with
zero RPO and zero RTO to survive if there is a disaster situation.
Cascaded
SRDF is a
three-way
data
mirroring
topology
providing a
zero data
loss solution
at long
distances
when there is
a primary
site failure. Cascaded SRDF configuration.
SRDF/Star Configuration
SRDF/Star operates in a
concurrent and cascaded
configuration:
• Concurrent SRDF/Star: Data
is mirrored from the primary
site concurrently to two R2
devices. Both the secondary
and tertiary sites are the
potential recovery sites if the
primary site fails.
• Cascaded SRDF/Star: Data is
mirrored first from the primary
site to a secondary site, and
then from a secondary site to a
tertiary site. Both the SRDF/Star configuration.
secondary and tertiary sites
are potential recovery sites.
Four-Site Configuration
Knowledge Check 1
a. Concurrent SRDF
b. Cascaded SRDF
c. SRDF/Star
d. Cascaded SRDF/Star
Knowledge Check 2
2. Read through the scenario. What solution could Jeff provide to the
Customer Engineer?
SRDF/S Introduction
• For read hits, data is read from cache on the primary array and sent to
the host.
• For read misses, data is read from the R1 device and sent to the host.
The symcfg command is used to list the remote adapters and RDF
groups. These operations include:
• Viewing the arrays and its details using the list operation.
• Displaying the remote adapters and the ports used using the list -
ra all operation.
• Displaying the RDF groups and its related information using the list
-rdfg all operation.
The symsan command helps in situations where SRDF groups have not
yet been created between the storage array pairs.
View the remote adapters, the RDF group, and the SRDF connectivity
using the SYMCLI interaction below. The environment for this interaction
includes:
The command symdev list pd displays the list of all the devices that
the host can access on the array.
This command correlates the host physical device name with the array
device number.
The physical device names from the host are used to format the devices,
create partitions, and create and mount file systems.
Create SRDF Group option and Create SRDF group dialog in Unisphere for
PowerMax.
To create dynamic SRDF groups for a selected array, use the Data
Protection menu option.
• Click Create SRDF Group to open the Create SRDF group dialog.
• Select the remote array and provide an SRDF group label.
• Select the SRDF group number and the local ports.
• Select the SRDF group number and the remote ports.
Use the simulation to create a dynamic SRDF group using Unisphere for
PowerMax.
The physical links and communication between the two arrays must exist
for this command to succeed.
Create the dynamic SRDF group using the SYMCLI interaction below. The
environment for this simulation includes:
Create Pairs option and Create Pairs dialog in Unisphere for PowerMax.
Use the simulation to create dynamic SRDF pairs using Unisphere for
PowerMax.
• Select the SRDF group and view the volumes associated with the
SRDF group.
• Select and delete all the volumes associated with the local and remote
array.
• Delete the SRDF group after all the SRDF group volumes are deleted.
• Array: 0639
• Device File: sess.txt
• SRDF Group: 201
Failover
A failover operation is a process that protects the data on the primary site
by moving production applications from the primary site to the secondary
site.
Failover option and dialog for an SRDF storage group in Unisphere for PowerMax.
• Select the storage group and use the Failover option from the more
actions menu.
• Select Type.
• Select Use Immediate with the operation to immediately drop the
SRDF/A session and make the devices Not Ready on the link.
Switch production to the target array using the SYMCLI simulation below.
The environment for this simulation includes:
Update
Performing failback right after failover resumes production on the R1. New
writes to the R1 have to be transferred synchronously to the R2.
Any track requested on the R1 that has not yet been transferred from the
R2 has to be read from across the links. This could lead to performance
degradation on the R1 devices.
R1 Update option and dialog for an SRDF storage group in Unisphere for PowerMax.
• Select the storage group and use the R1 Update option within Set
Volume Attributes from the More Actions menu.
Failback
Failback option and dialog for an SRDF storage group in Unisphere for PowerMax.
• Select the storage group and use the Failback option from the more
actions menu to invoke a failback operation.
Switch production to the source array using the SYMCLI simulation below.
The environment for this simulation includes:
During decision support operations, both the source and target sites are
healthy and available. The decision support operations for SRDF devices
include:
• Split an SRDF pair which stops mirroring for the SRDF devices in a
device pair.
• Establish—full or incremental—an SRDF pair by initiating a data copy
from the source volume to the target volume.
• Restore—full or incremental—remote mirroring, which initiates a data
copy from the target volume to the source volume.
Split
Split option and dialog for an SRDF storage group in Unisphere for PowerMax.
Split an SRDF pair using the SYMCLI simulation below. The environment
for this simulation includes:
Establish
Establish option and dialog for an SRDF storage group in Unisphere for PowerMax.
• Select the storage group and use the Establish option to invoke the
establish operation.
• Select Type – Incremental or full.
Restore
Restore option and dialog for an SRDF storage group in Unisphere for PowerMax.
SRDF/A Overview
SRDF/Asynchronous Replication.
Data in a delta set is processed using four cycle types. The table shows
cycle types and associated functions.
Apply cycle Changed blocks in the delta set are marked as invalid
tracks which are then destaged to the disks.
Multi-cycle mode (MCM) places host writes into cycles or chunks and
transfers entire chunks to the target system.
• Scenarios in which SRDF/A operates in MCM are:
− Arrays in the configuration run PowerMaxOS or HYPERMAX OS.
• MCM permits multiple cycles on the R1 side and only two cycles on the
R2 side.
− R1 side – one capture and one or more transmit.
− R2 side – one receive and one apply.
• MCM is the default for both SSC and MSC sessions.
• MCM improves the robustness of the SRDF/A session and reduces
spillover into the Delta Set Extension (DSE) storage pool.
Movie:
Overview
• Viewing the operational mode for the device group using the query
operation.
• Setting the mode of operation to Asynchronous using the set mode
async operation.
• Enabling the consistency using the enable operation.
Overview
Dynamically add and remove device pairs from an active SRDF/A session
using the consistency exempt option.
• When the consistency exempt flag is enabled, devices are excluded
from the consistency check on the group.
• The consistency exempt state is automatically terminated once the
operation is complete.
• The -exempt option can be used with the createpair, movepair,
and suspend operations.
• Devices that are suspended and consistency exempt can be controlled
independently of the other devices in the active SRDF/A session.
1. Suspend the device pair in the active SRDF/A group using the
suspend operation with the -exempt flag.
2. Move the device pair from the active SRDF/A group using the
movepair operation.
3. Resume the device pair using the resume operation.
4. Monitor synchronization using the verify -synchronized
operation.
SRDF/A Limitations
SRDF/A Delta Set Extension (DSE) extends the cache space available for
SRDF/A session cycles by offloading cycle data from cache to a
preconfigured storage pool.
• DSE offloads the delta set data when the system approaches the
maximum SRDF/A cache utilization threshold.
• DSE is enabled by default and can be configured independently on the
source and target arrays. Dell recommends using the same
configuration of DSE on both sides.
• Arrays are preconfigured with one or more Storage Resource Pools
(SRPs) before installation.
• The SRP designated for DSE supports the DSE allocations for all
SRDF/A sessions on the array.
− The default SRP for DSE is the default SRP for Fixed Block
Architecture (FBA) devices.
• Data is paged to disk when the array Write Pending count crosses the
DSE threshold.
− The default DSE threshold is 50% of the array Write Pending limit.
− When conditions return to normal, data is read back from disk to
cache and transmitted to the target array.
To set SRDF/A controls for SRDF storage groups on a selected array, use
the Protection dashboard.
Set SRDF/A option and dialog for an SRDF storage group in Unisphere for PowerMax.
• Select the SRDF storage group and from the More Actions menu use
the Set SRDF/A option from Asynchronous.
• Activate SRDF/A.
• Set Activate Type.
• Viewing the detailed list of all the configured SRPs using the list -
srp -detail operation.
• Viewing the details about specific SRP using the show -srp
operation.
• Modifying reserved capacity on the SRP using the set -resv_cap
operation.
Set the DSE SRP capacity using the SYMCLI simulation below. The
environment for this simulation includes:
Transmit Idle
Viewing the state of the device pairs using the -rdfg query operation
and use the -rdfa option for specific SRDF/A information.
Check the status of the SRDF device pair using the SYMCLI simulation
below. The environment for this simulation includes:
When host I/O rates spike or link slowdowns make the transmit or apply
cycle times longer, group pacing extends the host write I/O response time.
Response time is extended through RA group-level pacing. This
parameter adjusts the pace of incoming host I/O writes to match the SRDF
link bandwidth and transfer rate. Group-level write pacing is user enabled
or disabled.
Array-Wide Attributes
− DSE offloads some or all the delta set data into the SRP based on
the maximum SRDF/A cache utilization.
− Values range from 0 GB to 100000 GB. The default value is
NoLimit.
• Viewing the current settings of the array using the list -v operation.
• Setting DSE maximum capacity using the set -dse_max_cap
operation.
• Setting maximum cache percentage using the set -
rdfa_cache_percent operation.
• Setting time for throttling the host writes using the set -
rdfa_host_throttle_time operation.
Set the SRDF/A array-wide attributes using the SYMCLI simulation below.
The environment for this simulation includes:
Group Attributes
To set SRDF/A group attributes for SRDF groups on a selected array, use
the Data Protection option.
• Select the SRDF group and use the SRDF/A Setting option from the
More Actions menu.
• Set the Minimum Cycle Time.
• Set the Session Priority.
• Enable Transmit Idle.
• Enable Use Both Sides to set the attributes on both the source and
target arrays.
The symcfg command is used to list attributes and the symrdf command
is used to set SRDF/A group attributes.
• Viewing the group attributes of the array using the list -rdfa
operation.
• Setting minimum cycle time using the set rdfa -cycle_time
operation.
• Setting session priority using the set rdfa -priority operation.
• Enabling Transmit Idle using the set rdfa -transmit_idle
operation.
• Setting attributes on both local and remote arrays using the -
both_sides operation.
Set the SRDF/A group attributes using the SYMCLI simulation below. The
environment for this simulation includes:
Link Loss
SRDF links can be lost for more time than link limbo or Transmit Idle can
manage. During these extended link losses, all devices in the SRDF group
are set to a Not Ready state.
1The surge is due to shipping accumulated invalid tracks, and adding new
data with host writes to the SRDF/A cycles. This added data could lead to
SRDF/A consuming excessive amounts of cache and reaching the System
Write Pending Limit.
View the state of the device pairs using the -rdfg query operation with
the -rdfa option for specific SRDF/A information.
Check the state of the SRDF/A device pair after a link loss using the
SYMCLI simulation below. The environment for this simulation includes:
In this scenario, the links are restored and the pair state moves to
Suspended. Although the flags indicate SRDF/A mode, the session status
is Inactive. The R2 data is consistent because the data would be
consistent up to the last apply cycle. However, there are accumulated R2
invalid tracks that are owed to the R2 side.
View the state of the device pairs using the -rdfg query operation with
the -rdfa option for specific SRDF/A information.
Check the state of the SRDF/A device pair after the links are restored
using the SYMCLI simulation below. The environment for this simulation
includes:
The symrdf command is used to recover from a failed link. The steps to
perform this operation include:
1. As consistency was enabled when the links were lost, disable
consistency using the disable operation.
2. Resume the links using the resume operation.
View the state of the device pairs using the -rdfg query operation with
the -rdfa option for specific SRDF/A information.
Check the state of the SRDF/A device pair after resuming using the
SYMCLI simulation below. The environment for this simulation includes:
View the state of the device pairs using the query operation with the -
rdfa option for specific SRDF/A information.
Check the status of the SRDF device pair using the SYMCLI simulation
below. The environment for this simulation includes:
Failover
Failback
Overview
• If a session has devices from multiple arrays, the host running storrdfd
must have a connection to all the arrays to coordinate cycle switches.
• Dell Technologies recommends having multiple hosts running the
storrdfd daemon with connection to the arrays.
− If one host fails, the surviving host can continue processing MSC
cycle switches.
RDF Daemon
The daemon monitors for any failure to propagate data to the secondary
array devices and drops all SRDF/A sessions together to maintain
dependent-write consistency.
SRDF/A MSC operations are not active until storrdfd is enabled and
started in the SYMAPI options file.
The steps to start SRDF/A operations with Multi Session Consistency are:
1. Set the SYMAPI_USE_RDFD = ENABLE parameter in the options
configuration file.
2. Create a Composite Group (CG).
− Group definition is passed to the SRDF daemon as a candidate
group.
− The storrdfd daemon is started automatically.
3. Add all the devices in the SRDF/A sessions to the CG.
4. Place all the CG devices into Async mode.
5. Enable the CG.
6. When the devices become Read/Write enabled on the link, the
daemon:
1. Create a Composite Group (CG) using the create operation and the
-rdf_consistency option.
2. Add devices to the CG using the add operation.
3. View information about CG using the list operation.
4. View the properties of the CG using the show operation.
5. Enable consistency using the enable operation.
Manage two SRDF/A groups as a single entity using MSC using the
SYMCLI simulation below. The environment for this simulation includes:
SRDF/Metro Overview
SRDF/Metro Introduction
SRDF/Metro
delivers active-
active high
availability for non-
stop data access
and workload
mobility.
In SRDF/Metro,
both R1 and R2
devices are
Read/Write
accessible to the
host systems
simultaneously.
SRDF/Metro
synchronously
copies data that is
written to a device
SRDF/Metro configuration.
to another device in
the pair. This
ensures that both devices have identical content.
When the
devices in an
SRDF/Metro
group become
Ready on the
SRDF link,
PowerMaxOS
marks the
array that
contains the
R1 devices as
the bias. The
other array
becomes the
non-bias side.
SRDF/Metro resiliency configuration with no witness configured.
If a device pair
becomes Not Ready on the SRDF link, PowerMaxOS allows all devices
on the bias side to remain accessible to the application host. The devices
on the non-bias side become inaccessible to the application host.
The ActiveBias pair state indicates that the devices operating with device
bias are ready to provide high availability.
Up to 32 vWitnesses can be
configured for redundancy.
If two separate fault domains are unavailable for vWitness, the two
vWitnesses can be placed on a third fault domain but on a different Linux
or ESX server.
Disaster Recovery in an
SRDF/Metro
configuration can also be
accomplished using
SRDF/A from the
secondary site.
Primary Site with SRDF/A and Secondary Site with SRDF/A to Tertiary Site
The primary and secondary sites can also be protected with SRDF/A to
their respective tertiary sites.
SRDF/Metro Smart DR
SRDF/Metro provides
higher database resiliency
at short distances of up to
200 kilometers (125 miles)
based on synchronous
replication. To provide
longer distance resiliency,
SRDF/A replication is
used.
SRDF/Metro Management
− Create pair and move pair operations can add devices to the
SRDF/Metro group.
− Delete pair and move pair operations can remove devices from the
SRDF/Metro group.
Use the simulation to view the SRDF/Metro group details using Unisphere
for PowerMax.
SRDF/Metro Restrictions
Create Pairs option and Create Pairs dialog in Unisphere for PowerMax.
• Mirror Type – R1 or R2
• SRDF Mode – Active for SRDF/Metro
Moving SRDF pairs moves SRDF pairs from one SRDF group to another.
The move type can be a full move or a half move. A half move specifies to
move only the local half of the SRDF pair.
Use the simulation to move SRDF/Metro device pairs using Unisphere for
PowerMax.
SRDF/Metro Implementation
A physical witness for SRDF/Metro is an SRDF group that does not send
any data and serves the role of a third site to SRDF/Metro cluster for
arbitration purposes only. To create a physical witness group, create an
SRDF group from both the local and remote storage systems to a third
PowerMax storage system.
To create physical witness SRDF group on a selected array, use the Data
Protection menu option for SRDF groups.
• Select the communication protocol, remote array ID, and provide a
label for the SRDF group.
− Select the SRDF/Metro Witness Group checkbox to create a
physical witness SRDF group.
• Select the SRDF group number and select the local ports.
• Select the SRDF group number and select the remote ports.
Use the simulation to create physical witness group using Unisphere for
PowerMax.
The process of creaing physical witness groups must be done on both the
R1 and R2 sides of the Metro configuration. In this simulation, run through
the R1 side configuration.
Virtual Witness
Steps to protect storage group with SRDF/Metro using Unisphere for PowerMax.
Expand option and dialog for storage group volume in Unisphere for PowerMax.
Suspend action is issued for a configuration with no witness array during planned
SRDF outage.
In the example shown in the image, bias is used since there is no witness
array. The R1 devices are Read/Write to the host, and the R2 devices are
Not Ready to the host.
Once the planned outage is complete, the user can decide to keep the R1
or R2 data.
• The R2 which is the side that is Read/Write to the host, becomes the
R1.
• For a planned application outage to the R1 devices, which is the
default bias side, a half swap to both the R1 and R2 is allowed.
Note: Half swaps force a bias change when the devices are Not Ready
on the SRDF link.
Operations continue on
the R1 side, and the R2
side is Not Ready to the
host.
If the application
continues running on
the R1 side, the SRDF
pair state changes
from Partitioned to
Suspended once the
SRDF link failure is
addressed. At this
point, an establish or
restore operation can
be performed.
When the SRDF link failure is resolved, the device pair status remains
Partitioned because there are two R1s in the Metro configuration.
• There is an R1 and
R2 in the
configuration.
• Bias is applied on
the R1 side.
• The device pairs go
into a Suspended
state.
• The application can
now be brought Half swap is issued to the original R2.
online to the original
R1 side and either an establish or restore operation can be performed.
The configuration returns to its original condition.
Use the simulation to perfom the suspend and establish operations using
Unisphere for PowerMax when there is a planned outage.
Click the Save Progress & Exit button in the course menu or below
to record this content as complete.
Go to the next learning or assessment, if applicable.
ALUA
ALUA is a standard which defines how a host can access a storage device
through multiple paths. ALUA allows the host to use the most efficient path
to the storage device.
BCV
BCV refers to Business Continuity Volume which is a full copy of the
source device and can be mounted to the same or different host.
BFSI Sectors
BFSI sectors refers to Banking, Financial Services, and Insurance sectors.
Half Swap
Half swap operation swaps one half of an SRDF relationship to change an
R1 to R2 or R2 to R1. Half swap is performed when one end of the
devices are not available and a full personality swap cannot be performed.
Link Limbo
Link limbo allows you to set a specific length of time for Enginuity to wait
when a link goes down before updating the link status. Link limbo value
can be specified on the local side or both the local and remote sides. If the
link status is still not ready after the link limbo time expires, devices are
marked not ready to the link. The value of the link limbo timer can be 0
through 120 s. The default is 10 s. To protect from session drops after the
maximum link limbo time expires, the Transmit Idle feature may be
enabled.
Logical Device
Zero RPO
Zero Recovery Point Objective (RPO) is a setup in which no data loss is
accepatble in the event of an outage.
Zero RTO
Zero Recovery Time Objective (RTO) is a setup in which application
downtime is never acceptable when an outage occurs.