IBM GDPS V3.3: Improving Disaster Recovery Capabilities to Help Ensure a Highly Available, Resilient Business Environment
Overview
GDPS(TM) is IBM's premier continuous availability and disaster recovery solution. IBM is proud to announce the general availability of GDPS V3.3. Available on January 25, 2006, GDPS V3.3 offers:
Enhanced availability with autonomic detection of "Soft Failures" on disk control units to trigger a HyperSwap(TM)
Exploitation of XRC enhancements for increased scalability in large I/O configurations and configurations with intensive I/O characteristics
Ease of use to support z/OS® V7 XRC+ staging data sets
Expanded functionality to provide data consistency between disk and duplexed Coupling Facility (CF) structures
In addition, IBM is reannouncing the general availability of GDPS/Global Mirror (GDPS/GM). Based upon IBM TotalStorage® Global Mirror technology, IBM GDPS/Global Mirror automation can help simplify data replication across any number of IBM System z(TM) and/or open system servers to a remote site that can be at virtually any distance from the primary site. This can help ensure rapid recovery and restart capability for your IBM System z9(TM), zSeries®, and open systems data for testing purposes as well as both planned and unplanned outages. GDPS/GM also provides automation facilities to reconfigure your System z9 and zSeries servers and to restart the systems that run on these servers for testing and for actual disaster recovery.
GDPS/Global Mirror automation technology is designed to manage the IBM TotalStorage Global Mirror copy technology, monitor the mirroring configuration, and automate management and recovery tasks.
GDPS is also providing a new "three-site" solution combining the benefits of GDPS/PPRC using Metro Mirror with GDPS/Global Mirror using Global Mirror technology. This solution, GDPS Metro/Global Mirror, is designed to provide the near-continuous availability aspects of HyperSwap and help prevent data loss within the Metro Mirror environment, along with providing a long-distance disaster recovery solution with no response-time impact. Metro/Global Mirror has been available via an RPQ since October 31, 2005.
More detailed information on the GDPS service offerings is available on the Internet at
http://www.ibm.com/servers/eserver/zseries/gdps
Availability date
Available now (as of January 25, 2006):
RCMF/PPRC V3.3
GDPS/PPRC V3.3
GDPS/PPRC HyperSwap Manager V3.3
RCMF/XRC V3.3
GDPS/XRC V3.3
GDPS/Global Mirror V3.3
GDPS Metro/Global Mirror V3.3
Description
IBM Global Services continues to enhance GDPS with:
Extended HyperSwap functionality with IOS timing trigger
Improved availability with enhanced recovery support in a CF structure duplexing environment
Performance improvements for System Logger in a z/OS Global Mirror (previously known as XRC) environment
Scalability improvements for XRC
Unlimited distance solution for z/OS and open data with the new GDPS/Global Mirror offering
Unplanned HyperSwap IOS timing trigger
If a disk subsystem experiences a "hard failure" such as a boxed device, rank array failure, or disk subsystem failure, current versions of GDPS/PPRC and GDPS/PPRC HyperSwap Manager (GDPS/PPRC HM) are designed to detect this and automatically invoke HyperSwap to transparently switch all primary PPRC disks with the secondary disks within seconds.
Occasionally, no signal comes back after an I/O operation has started. The I/O starts, but it is as if it doesn't end. There are no errors returned. The only indication that something is wrong is that the z/OS I/O Missing Interrupt Handler (MIH) detects this and generates a message. It is then up to the operator to see the message and figure out what to do. By that time, it is possible that the transactions waiting for I/O and holding on to resources can in turn cause other transactions to wait, and can bring the entire system to a stop.
The HyperSwap IOS timing trigger is designed to allow HyperSwap to be invoked automatically when user-defined I/O timing thresholds are exceeded. In a matter of seconds, transactions can now resume processing on the secondary disk, providing availability benefits and avoiding operator intervention.
The HyperSwap IOS Timing trigger requires APAR OA11750 available on z/OS V1.4.
HyperSwap is available with the GDPS/PPRC and GDPS/PPRC HyperSwap Manager offerings.
GDPS enhanced recovery support
In the event of a primary site failure, the current GDPS/PPRC cannot ensure that the CF structure data may be time-consistent with the "frozen" copy of data on disk, so GDPS must discard all CF structures at the secondary site when restarting workloads. This results in loss of "changed" data in CF structures. Users must execute potentially long-running and highly variable data recovery procedures to restore the lost CF data.
GDPS enhanced recovery is designed to ensure that the secondary PPRC volumes and the CF structures are time consistent, thereby helping to provide consistent application restart times without any special recovery procedures.
If you specify the FREEZE=STOP policy with GDPS/PPRC and duplex the appropriate CF structures, when CF structure duplexing drops into simplex, GDPS is designed to direct z/OS to always keep the CF structures in the site where the secondary disks reside. This helps to insure the PPRC volumes and recovery-site CF structures are time consistent thereby providing consistent application restart times without any special recovery procedures. This is especially significant for customers using DB2® data sharing, IMS(TM) with shared DEDB/VSO, or WebSphere® MQ shared queues.
GDPS enhanced recovery support requires z/OS APAR OA11719, available back to z/OS V1.5.
Improving performance
System logger provides new support for XRC+ by allowing you to choose asynchronous writes to staging data sets for logstreams. Previously, all writes had to be synchronous. This limited the throughput for high-volume logging applications such as WebSphere, CICS®, and IMS. The ability to do asynchronous writes can allow the use of z/OS Global Mirror (XRC) for some applications for which it was not previously practical. XRC+ is available on z/OS and z/OS.e V1.7.
Refer to Preview: IBM z/OS V1.7 and z/OS.e V1.7: World-class computing for On Demand Business, Software Announcement 205-034 dated February 15, 2005.
GDPS/XRC has extended its automation to support XRC+. It is designed to provide the ability to configure and manage the staging data set remote copy pairs.
Scalability
GDPS/XRC support is being extended to help improve XRC scalability for large systems by:
Write Pacing
APAR OA09239 provides for the new XRC Write Pacing support. By automatically inserting delays into the I/O response for high-intensity update applications, XRC can then prevent the secondary disk in the remote site from falling behind, delaying the RPO for all applications.
Exploitation of the Write Pacing function on GDPS/XRC systems requires APAR 65 (AG31D65), which is fully compatible with all existing supported GDPS/XRC software levels.
Parallel execution
Previously, GDPS typically processed all XRC System Data Movers (SDMs) in sequence within an LPAR. With GDPS V3.3, many XRC commands can now be done in parallel across all the SDMs. A parallel execution of XRC commands across all SDMs allows for improved responsiveness, improved usability, and reduced recovery time.
Support for more than 14 SDMs
Previously, XRC supported up to 14 coupled SDMs, split across up to five SDM address spaces per z/OS LPAR. New support expands this to allow up to 14 Coupled eXtended Remote Copy (CXRC) sessions. Each CXRC can consist of one or more XRC logical sessions. Additionally, Multiple eXtended Remote Copy (MXRC) currently allows the user to run up to five XRC logical sessions within a single LPAR. This enhancement will allow significantly more SDMs, thereby increasing the number of parallel paths to transfer data. This allows GDPS/XRC to handle larger configurations and higher throughputs while maintaining the client's service level agreements. More information on CXRC can be found in z/OS DFSMS Advanced Copy Services (SC35-0428-09).
The planned availability of GDPS support for more than 14 coupled SDMs is second quarter 2006.
XRC Performance Monitor (XPM) updates
In addition to the above enhancements, XPM is being modified to support the new larger master sessions. XPM will have the ability to display (via the Interactive Interface) and process (via the Exception Batch Monitor) cluster-level data. The Interactive Interface will be modified to recognize and display consolidated cluster data and larger values for data-movement-related statistics. The planned availability of the XPM updates is March 31, 2006.
GDPS V3.3 is available as of January 25, 2006. GDPS is designed to work in conjunction with the z9-109, z990, z890, z900, and z800 servers. For a complete list of other supported hardware platforms and software prerequisites, refer to the GDPS Web site
http://www.ibm.com/server/eserver/zseries/gdps
GDPS/Global Mirror has been available as of October 2005. Contact your IBM representative or send an e-mail to GDPS@us.ibm.com for information regarding ordering GDPS.
GDPS/Global Mirror was previewed in IBM zSeries 990 and 890 FICON(TM) enhancements Hardware Announcement 105-012 , dated January 25, 2005.
Accessibility by people with disabilities
A U.S. Section 508 Voluntary Product Accessibility Template (VPAT) containing details on the product's accessibility compliance can be requested via IBM's Web site
http://3.ibm.com/able/product_accessibility/index.html
Product positioning
The GDPS solution suite includes six different service offerings to meet different customer requirements:
RCMF/PPRC
Remote Copy Management Facility (RCMF) provides management of the remote copy environment and disk configuration from a central point of control. The RCMF/PPRC offering can be used to manage a PPRC (Metro Mirror) remote copy environment.
RCMF/XRC
RCMF/XRC is a disaster recovery offering which can be used to manage an XRC (z/OS Global Mirror) remote copy environment.
GDPS/PPRC HyperSwap Manager
GDPS/PPRC HyperSwap Manager provides either a single-site near-continuous availability solution or a multi-site disaster recovery solution. It is an entry-level solution available at a cost-effective price. GDPS/PPRC HyperSwap Manager is designed to allow customers to increase availability and provide applications with continuous access to data. Today, GDPS/PPRC HyperSwap Manager appeals to zSeries customers who require continuous availability and extremely fast recovery.
Within a single site, or multiple sites, GDPS/PPRC HyperSwap Manager extends Parallel Sysplex® availability to disk subsystems by masking planned and unplanned disk outages caused by disk maintenance and disk failures. It also provides management of the data replication environment and automates switching between the two copies of the data without causing an application outage, therefore providing near-continuous access to data.
The GDPS/PPRC HyperSwap Manager solution is a subset of the full GDPS/PPRC solution, designed to provide a very affordable entry point to the full family of GDPS/PPRC offerings. It features specially priced limited-function Tivoli® System Automation and NetView® software products, thus satisfying the GDPS software automation prerequisites with a lower price and a cost-effective entry point to the GDPS family of offerings. Users who already have the full-function Tivoli System Automation and NetView software products may continue to use them as the prerequisites for GDPS/PPRC HyperSwap Manager.
A customer can migrate from a GDPS/PPRC HyperSwap Manager implementation to the full-function GDPS/PPRC capability as business requirements demand shorter recovery time objectives. The initial investment in GDPS/PPRC HyperSwap Manager is protected when you choose to move to full-function GDPS/PPRC by leveraging the existing GDPS/PPRC HyperSwap Manager implementation and skills.
GDPS/PPRC
GDPS/PPRC complements a multisite Parallel Sysplex implementation by providing a single, automated solution to dynamically manage storage subsystem mirroring, disk and tape, processors, and network resources. It is designed to help a business attain continuous availability and near-transparent business continuity (disaster recovery) with data consistency and no or minimal data loss. GDPS/PPRC is designed to minimize and potentially eliminate the impact of any failure, including disasters, or a planned outage.
GDPS/PPRC is a full-function offering that includes the capabilities of GDPS/PPRC HM. It is designed to provide an automated end-to-end solution to dynamically manage storage system mirroring, processors, and network resources for planned and unplanned events that could interrupt continued IT business operations.
The GDPS/PPRC offering is a world-class solution built on the z/OS platform and yet can manage a heterogeneous environment.
GDPS/PPRC is designed to provide the ability to perform a controlled site switch for both planned and unplanned site outages, with no or minimal data loss, maintaining full data integrity across multiple volumes and storage subsystems and the ability to perform a normal Data Base Management System (DBMS) restart - not DBMS recovery - in the second site. GDPS/PPRC is application-independent and therefore can cover your complete application environment.
GDPS/XRC
Based upon IBM TotalStorage z/OS Global Mirror (Extended Remote Copy, or XRC), GDPS/XRC is a combined hardware and z/OS software asynchronous remote-copy solution. Consistency of the data is maintained via the Consistency Group function within the System Data Mover. GDPS/XRC includes automation to manage remote copy pairs and automates the process of recovering the production environment with limited manual intervention, including invocation of CBU, thus providing significant value in reducing the duration of the recovery window and requiring less operator interaction. GDPS/XRC is capable of the following attributes:
Disaster recovery solution
RTO between an hour to two hours
RPO less than one minute
Protects against localized or regional disasters, depending on the distance between the application site and the disaster recovery site (distance between sites is unlimited)
Minimal remote copy performance impact
GDPS/XRC is well suited for large System z workloads and can be used for business continuance solutions, workload movement, and data migration.
Because of the asynchronous nature of XRC, it is possible to have the secondary disk at greater distances than would be acceptable for Metro Mirror (synchronous PPRC). Channel extender technology can be used to place the secondary disk thousands of kilometers away.
In some cases an asynchronous disaster recovery solution is more desirable than one that uses synchronous technology. Sometimes applications are too sensitive to accept the additional latency incurred when using synchronous copy technology.
GDPS/Global Mirror
The latest member of the GDPS suite of offerings, GDPS/Global Mirror offers a multisite, end-to-end disaster recovery solution for your IBM z/OS systems and open systems data.
IBM GDPS/Global Mirror automation technology can help simplify data replication across any number of System z(TM) systems and/or open system servers to a remote site that can be at virtually any distance from the primary site. This can help ensure rapid recovery and restart capability of your environment for both testing and disaster recovery, and restart capability for your open systems environment for testing and disaster recovery. Being able to test and practice recovery allows you to build skills in order to be ready when a disaster occurs.
GDPS/Global Mirror automation technology is designed to manage the IBM TotalStorage Global Mirror copy services and the disk configuration, monitor the mirroring environment, and automate management and recovery tasks. It can perform failure recovery from a central point of control. This can provide the ability to synchronize System z and open systems data at virtually any distance from your primary site.
The point-in-time copy functionality offered by the IBM TotalStorage Global Mirror technology allows you to initiate a restart of your database managers on any supported platform, to help reduce complexity and avoid having to create and maintain different recovery procedures for each of your database managers.
All this helps provide a comprehensive disaster recovery solution.
The six offerings listed above can be combined as follows:
GDPS/PPRC used with GDPS/XRC (GDPS PPRC/XRC)
GDPS PPRC/XRC provides the ability to combine the advantages of metropolitan-distance Business Continuity and regional or long-distance Disaster Recovery. This can provide a near-continuous availability solution with no data loss and minimum application impact across two sites located at metropolitan distances, and a disaster recovery solution with recovery at an out-of-region site with minimal data loss.
A typical GDPS PPRC/XRC configuration has the primary disk copying data synchronously to a location within the metropolitan area using Metro Mirror (PPRC), as well as asynchronously to a remote disk subsystem a long distance away via z/OS Global Mirror (XRC). This enables a z/OS three-site high availability and disaster recovery solution for even greater protection from planned and unplanned outages.
Combining the benefits of PPRC and XRC, GDPS PPRC/XRC enables:
HyperSwap capability for near-continuous availability for a disk control unit failure
Option for no data loss
Data consistency to allow restart, not recovery
Long-distance disaster recovery site for protection against a regional disaster
Minimal application impact
GDPS automation to manage remote copy pairs, manage a Parallel Sysplex configuration, and perform planned as well as unplanned reconfigurations
The same primary volume is used for both PPRC and XRC data replication and can support two different GDPSs: GDPS/PPRC for metropolitan distance and business continuity, and GDPS/XRC for regional distance and disaster recovery.
The two mirroring technologies and GDPS implementations work independently of each other, yet provide the synergy of a common management scheme and common skills.
Since GDPS/XRC supports zSeries data only (z/OS, Linux on zSeries), GDPS XRC is a zSeries solution only.
GDPS/PPRC used with GDPS/Global Mirror (GDPS Metro/Global Mirror)
GDPS Metro/Global Mirror has the benefit of being able to manage across the configuration all formats of data, as Global Mirror is not limited to zSeries formatted data.
GDPS Metro/Global Mirror combines the benefits of GDPS/PPRC using Metro Mirror, with GDPS/Global Mirror using IBM TotalStorage Global Mirror. A typical configuration has the secondary disk from a Metro Mirror remote copy configuration in turn becoming the primary disk for a Global Mirror remote copy pair. Data is replicated in a "cascading" fashion.
Combining the benefits of PPRC and Global Mirror, GDPS Metro/Global Mirror enables:
HyperSwap capability for near-continuous availability for a disk control unit failure
Option for no data loss
Maintain disaster recovery capability after a HyperSwap
Data consistency to allow restart, not recovery, at either site 2 or site 3
Long-distance disaster recovery site for protection against a regional disaster
Minimal application impact
GDPS automation to manage remote copy pairs, manage a Parallel Sysplex configuration, and perform planned as well as unplanned reconfigurations
In addition, GDPS Metro/Global Mirror can do this for both zSeries as well as open data, and provide consistency between them.
GDPS Metro/Global Mirror is only available via RPQ.
Reference information
Enhancements to the IBM zSeries 900 Family of Servers, Hardware Announcement 101-308 , dated October 4, 2001
New Functions for IBM zSeries Servers Enhance Connectivity, Hardware Announcement 102-209 , dated August 13, 2002
IBM Introduces the IBM zSeries 990 Family of Servers, Hardware Announcement 103-142 , dated May 13, 2003
IBM enhances the IBM zSeries 990 family of servers, Hardware Announcement 103-280 , dated October 7, 2003
IBM Implementation Services, Installation Services, and Operational Support Services Now Available for Selected IBM Products, Services Announcement 603-015 , dated June 17, 2003
IBM TotalStorage PtP VTS includes FICON connectivity for increased performance and distance, Hardware Announcement 103-204 , dated July 15, 2004
IBM enhances the IBM zSeries 990 family of servers, Hardware Announcement 104-118 , dated April 7, 2004
Significant IBM zSeries mainframe security, SAN, and LAN innovations, Hardware Announcement 104-346 , dated October 7, 2004
IBM zSeries 990 and 890 FICON enhancements, Hardware Announcement 105-012 , dated January 25, 2005
Preview: IBM z/OS V1.7 and z/OS.e V1.7: World-class computing for On Demand Business, Software Announcement 205-034 , dated February 15, 2005
GDPS/PPRC HyperSwap Manager: Providing continuous availability of consistent data, Marketing Announcement 305-015 , dated February 15, 2005
IBM System z9 109 - The server built to protect and grow with your on demand enterprise, Hardware Announcement 105-241 , dated July 27, 2005
IBM Implementation Services for Geographically Dispersed Parallel Sysplex(TM) for managing disk mirroring using IBM Global Mirroring, Services Announcement 605-035 , dated October 18, 2005
Order now
To order, contact the Americas Call Centers or your local IBM representative.
To identify your local IBM representative, call 800-IBM-4YOU (426-4968).
Phone: 888-426-4343. (Select option for IBM
Service Offering.)
Internet: If you are an IBM Business Partner, sign on
to PartnerWorld. From Shortcuts, select
Online Technical Request.
The Americas Call Centers, our national direct marketing organization, can add your name to the mailing list for catalogs of IBM products.
Business Partner information
If you are a Direct Reseller - System Reseller acquiring products from IBM, you may link directly to Business Partner information for this announcement. A PartnerWorld ID and password are required (use IBM ID).
BP Attachment for Announcement Letter 306-024
