Home > Storage > PowerMax and VMAX > Storage Admin > Implementing Dell SRDF SRA with VMware SRM > SRDF/Star
SRDF Symmetrix Triangular Asynchronous Replication (SRDF/Star) is a data protection and failure recovery solution that covers three geographically dispersed data centers in a triangular topology.
SRDF/Star protects business data against a primary site failure or a regional disaster using concurrent SRDF or cascaded SRDF capabilities. These technologies mirror the same production data synchronously to a nearby remote site and asynchronously to a distant remote site. This architecture can be expanded to include multiple VMAX/PowerMax triangles.
Note: SRDF/Metro does not support SRDF/Star configurations.
SRDF/Star consists of three sites. The terminology that will be used throughout this book to refer to these three sites are:
SRDF/Star provides consistent data protection and incremental data recovery between target sites in the event of a workload site failure or transient fault (link failure). If a workload site becomes inoperable, SRDF/Star provides failover capability through SRDF/A recovery links that quickly reestablish data replication between the target sites. One target site resumes data replication for the workload site while the other resumes as a protected secondary target site. SRDF/Star requires SRDF/A recovery links and an additional control host at one of the target sites.
Each VMAX/PowerMax array in an SRDF/Star configuration uses dynamic SRDF devices that can function as either an R1 or an R2 device. During failure recovery, the R2 devices at either the synchronous target site or the asynchronous target site are dynamically converted to R1 devices to become production devices at the new workload site.
SRDF/Star data consistency is achieved through a composite group (CG) that leverages Multi-Session Consistency to ensure all members in the CG are either replicating or not replicating at any given point. A composite group enabled for MSC is referred to as an RDF consistency group. Devices in an RDF consistency group are protected to preserve the dependent-write consistency of a database, which may be distributed across multiple SRDF platforms.
Note: Since SRDF/Star environments are built on RDF consistency group technology, an RDF daemon must be running on at least one control host attached locally to each site. Dell strongly recommends running redundant RDF daemons on multiple control hosts to ensure that at least one RDF daemon is available to perform time-critical, consistency monitoring operations. By running redundant RDF daemons one avoids service interruptions caused by performance bottlenecks local to a control host, and link failures of the redundant RDF daemons and control hosts.
SRDF/Star environments provide advanced multi-site business continuity protection. They combine SRDF technologies to enable concurrent or cascaded SRDF/S and SRDF/A operations from the same source volumes. The SRDF/Star operation is a combination of host software and HYPERMAX OS/PowerMaxOS functionality that operates in a concurrent or cascaded SRDF configuration.
SRDF/Star provides the following benefits and features:
The SRDF SRA supports two types of SRDF/Star configurations: Concurrent SRDF/Star and Cascaded SRDF/Star. Diskless Cascaded SRDF/Star, otherwise known as SRDF/EDP (Extended Distance Protection) is not supported with the SRDF SRA. The two supported configurations are discussed briefly in the following subsections. For detailed information on setting up and configuring Star, refer to SRDF documentation.
Controlling SRDF/Star involves tasks such as bringing up the SRDF/Star sites for normal operation, isolating one or more sites for testing or other purposes, or switching the workload to one or more of the remote sites after workload site failure. Understanding these operations are integral to understanding the different operations performed by the SRDF SRA. Users can perform these and other SRDF/Star operations using the Solutions Enabler symstar command[5]. Common operations are listed and described in Table 1.
Name | Results |
Cleanup | Cleans up internal meta information and VMAX/PowerMax cache at the remote site after a failure at the workload site. |
Configure | Upgrades or transitions an existing SRDF/Star environment to employ R22 devices, provided the current SRDF/Star environment is operating in normal condition. |
Connect | Starts the SRDF data flow |
Disable | Disables SRDF/Star consistency protection across the three sites. |
Disconnect | Suspends the SRDF data flow and transitions the path to adaptive copy disk mode. |
Enable | Enables complete SRDF/Star consistency protection across the three sites. |
Halt | Used to prepare the system for a planned switch of the workload to a target site. This action write-disables the R1 devices, drains all invalid tracks and MSC cycles so that Site A=Site B=Site C, suspends SRDF links, disables all consistency protection, and sets adaptive copy disk mode. |
Isolate | Isolates one target site from the SRDF/Star configuration and makes its R2 devices read/write enabled to their hosts. |
List | Lists each SRDF/Star composite group configuration, including workload name, mode of operation, CG and Star states, and target names and states. |
Modifycg | Maintains consistency protection when adding or removing device pairs from an SRDF/Star consistency group. |
Protect | Synchronizes devices between the workload and target sites and enables SRDF/Star consistency protection to the specified target site. |
Query | Displays the status of a given SRDF/Star site configuration. |
Reconfigure | Transitions the SRDF/Star setup from concurrent SRDF to cascaded SRDF or vice versa after a site or link failure, or as part of a planned event. |
Reset | Cleans up internal meta information and VMAX/PowerMax cache at the remote site after transient fault (such as a loss of connectivity to the synchronous or asynchronous target site). |
Show | Displays the contents of the internal definition for a given SRDF/Star site configuration. |
Switch | Transitions workload operations to a target site after a workload site failure or as part of a planned event. |
Unprotect | Disables SRDF/Star consistency protection to the specified target site. |
Verify | Returns success if the state specified by the user matches the state of the star setup. |
In a concurrent SRDF configuration, data on a single source device is remotely mirrored to two target devices at the same time, providing two available copies of data. These mirrors operate independently but concurrently using any combination of SRDF modes.
Concurrent SRDF/Star environments provide a rapid reestablishment of replication operations in the event of a workload site failure. Rather than performing a full resynchronization between the asynchronous and synchronous target sites, concurrent SRDF/Star performs a differential synchronization, which dramatically reduces the time it takes to remotely protect the new workload site after the primary site failure. With Concurrent SRDF/Star one can determine which site contains the most current data in the event of a rolling disaster that affects the workload site. Figure 5 illustrates a Concurrent SRDF/Star configuration. The workload site (New York) transfers data synchronously to the target site (New Jersey) and asynchronously to the distant target site (London). The recovery links are between London and New Jersey, and if an outage happens at New York, an SRDF/A session can be established between these two target sites.
Figure 5. Concurrent SRDF Star example
Concurrent SRDF/Star is valuable for duplicate restarts and disaster recovery and provides increased flexibility for data mobility and application migrations.
Cascaded SRDF is a three-way data mirroring and recovery solution that provides enhanced replication capabilities, greater interoperability, and multiple ease-of-use improvements. A cascaded SRDF configuration does not require three separate site locations, although that is the most common configuration for a disaster recovery solution.
Cascaded SRDF introduces the concept of the dual role R1/R2 device, referred to as an R21 device. The R21 device is both an R1 mirror and an R2 mirror, for use only in cascaded SRDF operations. When thinking of the R21 device, it is easier to understand the concept if it is thought of as a mirror type, instead of as a device. The controls for these devices are relationship-based.
Figure 6 illustrates a Cascaded SRDF/Star configuration. In the event of a synchronous target site loss, a differential data resynchronization can be performed between the workload site and the asynchronous target site. This is achieved by using change tracker technology at the workload site to keep track of the data difference between the workload site and the asynchronous target site. In the case of Cascaded SRDF/Star, the synchronous target site is always more current than the asynchronous target site. In all cases, however, the choice of which site to use in the event of a failure is left to the discretion of the customer.
Figure 6. Cascaded SRDF/Star example
[5] Unisphere for PowerMax do not yet support most SRDF/Star operations.