IBM System Storage SAN Storage Solutions for System

IBM System Storage® Agenda § I 5/OS Specific Architecture § DS Copy Services §

IBM System Storage® I 5/OS Partition Single-Level Storage Main Memory © 2008 IBM Corporation

IBM System Storage® Auxiliary Storage Pools (Disk Pools) System ASP (Disk pool 1) User ASPs (Disk Pools 2 -255) Traditional User ASPs (Basic Pools 2 -32) Sysbas Independent ASPs - IASPs ( Independent Pools 33 -255) © 2008 IBM Corporation

IBM System Storage® IASP Attributes § IASP’s are “like” other mountable file systems on other platforms. § Must be part of an System i Cluster to be able to be switched/moved to a second system § Always requires work management changes to the operational environment to use this solution § May require some application changes, a few customers have found this. In house testing or an IASP feasibility study should be performed to check this. © 2008 IBM Corporation

IBM System Storage® © 2008 IBM Corporation

IBM System Storage® System i Journaling Your Program New Row Put File A Journal The DB is the last to know File A: PT New Row File A 2 1 Journal Receiver's content PT: New Row PT New Row Jrn Rcvr © 2008 IBM Corporation

IBM System Storage® Flash. Copy © 2008 IBM Corporation

IBM System Storage® Scenarios with Flash. Copy IASP Full System IASP Bf. S Flash. Copy Backup partition Bf. S • Cluster Production partition Flash. Copy Backup partition IASP Bf. S = Boot from SAN © 2008 IBM Corporation

IBM System Storage® Flashcopy - Benefits to the Customer § The production application downtime is only is as long as it takes to vary off the IASP, perform a Flash. Copy of the volumes in the IASP, and vary on IASP. Or, Power-down system, perform a Flash. Copy and IPL § On V 6 R 1 we can quiesce data to disk without power-down production system i § The performance impact on the production application during the save to tape operation is minimal since it is only influenced by the Flash. Copy activity which is mostly confined within the DS system. . § This solution can be implemented together with Backup, Recovery and Media Services for i 5/OS (BRMS) © 2008 IBM Corporation

IBM System Storage® Space Efficient Flash. Copy Production B f S Flash. Copy SE Backup Test at the residency: Workload CPW with 20 000 users, Production capacity 1125 GB, repository 280 GB Duration 4 h, Repository occupied about 25% © 2008 IBM Corporation

IBM System Storage® Consistency group and dependant writes in System i Without CG With CG Flashcopy 1. Journal Flashed 2. DB record § Journalling – Journal is first written to disk, then database record is written • If DB makes a change that crosses segment boundary it results i 3 writes: • first at the bottom of segment • second at the top of segment, and 3 write to update record was successful. • 3. write must follow writes 1 and 2. © 2008 IBM Corporation

IBM System Storage® Flash. Copy integration with BRMS Production System Backup System Enable Flash. Copy function for BRMS Set BRMS to backup Flash. Copy state CALL PGM(QBRM/Q 1 AOLD) PARM('FLASHSYS ' '*YES') Set BRMS to Flash. Copy mode CALL QBRM/Q 1 AOLD PARM(‘FLSSYSSTS’ ‘*BEGIN’) CALL QBRM/Q 1 AOLD PARM(‘FLSSYSSTS’ ‘*BACKUPSYS’) Backup Set BRMS to End backup mode QSYS/CALL QBRM/Q 1 AOLD PARM(‘FLSSYSSTS’ ‘*ENDBACKUP’) Send QUSRBRM to Production system © 2008 IBM Corporation

IBM System Storage® Quiesce data in an IASP or Sysbas Memory IASP, sysbas § To quiesce data in an IASP to disk use command CHGASPACT with the following parameters: – ASP Device (ASPDEV) specify the name of the IASP being quiesced, or *SYSBAS. – Option (OPTION) specify option *SUSPEND to quiesce the data in the IASP. – Suspend timeout (SSPTIMO) specify duration of the timeout during which the system is quescing the data to disk. – Suspend timeout action (SSPTIMOACN) specify the desired action at the end of timeout if the system was not able to quesce all data during the timeout. § Requires V 6 R 1 © 2008 IBM Corporation

IBM System Storage® Metro Mirror Suspend Bitmap Failover Bitmap Faiback or © 2008 IBM

IBM System Storage® Scenarios with Metro Mirror Full System Metro Mirror Bf. S System

IBM System Storage® Metro Mirror – Benefits to the Customer § When using IASP: Basically, recovery takes as long as is needed to vary on an IASP, Therefore recovery time is significantly shorter than with solutions which use Boot from SAN, or external Loadsource, where IPL is needed for recovery. § This solution has only impact on write operations performances. It doesn’t use any of the System i 5 CPU resources. § When using IASP: With this solution we can run a non production workload, like test or development tasks in the remote partition. This is again different from the implementation using Boot from SAN where the recovery partition has to be in standby, typically without any workload. § The System i Copy Services Toolkit which is used for this solution provides full automation of steps performed at planned or unplanned outages. All the needed steps for failover and failback are initiated by one command in a partition of System i. § Once this solution is established it requires very little maintenance and administration. Therefore some customers may prefer it to other high availability solutions available in System i. © 2008 IBM Corporation

IBM System Storage® Global Mirror Consistency group formation Data transmission 1 0 0 0 1 0 Global Copy 1 0 A B 0 1 0 C Consistency group save Create consistency group Flash. Copy issued with revertible option Drain consistency group and send to remote disk subsystem using Global Copy. Transmit updates in Global Copy mode Consistency group interval– 0 to 18 hrs Flash. Copy committed Start next consistency group © 2008 IBM Corporation

IBM System Storage® Global Mirror Recovery Some important steps: • Failover Global copy • IPL or vary-on IASP from B 4 • Reverse • Global Mirror 1 0 0 1 0 1 1 0 B 0 A 3 0 0 1 0 2 • Flashcopy to B C • Revert or commit Flashcopy if needed §The action to be taken depends on where in the cycle of creating a consistency §group we were at the time of the disaster event • Create CG Action Required • Drain CG • Issue FC • Commit FC • Action is based on observing Flashcopy status and sequence numbers © 2008 IBM Corporation

IBM System Storage® Scenarios with Global Mirror Full System Bf. S GM System i

IBM System Storage® Benefits to the Customer § Very small impact to production performances § Suitable solution for long distances § Customer has the possibility to balance between RPO and bandwidth § Due to consistency group on production volumes i 5/OS dependant writes are preserved § Once this solution is established it requires very little maintenance and administration § The System i Copy Services Toolkit is designed to do all neccesary steps at recovery autimatically. TPC-R can be used to do necessary steps at recovery on System Storage. © 2008 IBM Corporation

IBM System Storage® GM Recovery Point Objective and Recovery Time with System i § Recovery Point Objective (RPO) – Data on C volumes are 1 to 2 CG intervals behind data on A, depending at which point in CG interval the failure happened – Average RPO = 1. 5 CG interval § Recovery Time – GM of IASP: up to 30 minutes – GM with Boot from SAN: as long as it takes to IPL • Some measured durations of abnormal IPL: • I 570 V 5 R 3 – 44 min • I 570 V 5 R 4 – 1 h: 22 min © 2008 IBM Corporation

IBM System Storage® Customers with Global Mirror and System i § AAA Cooper Transportation, US – DS 8100, System i, Global Mirror, distance 2000 miles § Crete Carrier Corporation, US – DS 8000, system i, Global Mirror § Estes Express Lines, Inc. , US – DS 8100, ESS, System i, Global Mirror, distance 2000 miles § A Retail bank in UK – DS 8100, System i, distance 200 km § A customer in Italy – DS 6000, System i, Global Mirror, distance 30 km § A customer in Italy – DS 8000, System i, Global Mirror, distance 400 km § A customer in Czech republic – DS 8000, System i, Global Mirror, distance to US © 2008 IBM Corporation

IBM System Storage® Combined scenarios § Metro Mirror and Flash. Copy - Supported with System i – Supported by the Copy Services for System i § Global Mirror and Flash. Copy – Supported with System i – Supported by the Copy Services for System i § Metro Global Mirror – Supported with System i – On the roadmap for Copy Services for System i © 2008 IBM Corporation

IBM System Storage® What is Copy Services for System i § Services and program code for high availability and disaster recovery of System i. § The Copy Services for System i (Toolkit) contains code which gets installed on all partitions in a Flash. Copy, Metro Mirror or Global Mirror clustered environment; or management partition. § Employing System i clustering function with DS 8000 Copy Services. § Works with IASPs. § Works with full disk space. § Requires approval from Rochester to sell the solution. § Sold as service through the Lab Services - Client Technology Center. First year maintenance is required as part of the sale. § Works on V 6 R 1 and previous i 5/OS releases. © 2008 IBM Corporation

IBM System Storage® § Copy Services for System i - Full System Copy Services Flash. Copy LPAR DS 8000 § Power down the production partition § Flash. Copy the disks § Power up the production partition, return to operations LPAR Fl as h. C o py Production i. Series § Power up the Flash. Copy partition § Perform backups (or any other operation) on the Flash. Copy partition (normal backup time) System Storage Mirroring § Full System Flash. Copy Toolkit automates all of this for you ! § Planned: Quiesce data in *sysbas with CHGASPACT § Planned: Support for Metro Mirror and Global Mirror Backup i. Series Partition (powered off) © 2008 IBM Corporation

IBM System Storage® Example: Full System Flash. Copy © 2008 IBM Corporation

IBM System Storage® Copy Services for System i - IASP § Flash Copy, Metro or Global Mirror generates a second IASP copy identical to first § Changes to production IASP replicated to second copy of IASP § Cluster and SAN combined replication solution § Switch to DR site or activate Flashcopy is done with one command § IBM only HA solution OS and other system data I 5/OS cluster services IASP Copy Metro Mirror Flash. Copy DS 8000 © 2008 IBM Corporation

IBM System Storage® Release 6. 1 of Copy Services for System i IASP Toolkit Enhancements § Requires Power 5/6 and i 5/OS V 5 R 4 § Updated commands § WRKCSE enahncements – Scripts may be run intercatively from WRKCSE command – Volume count validation § Option in MM and GM to show out-of-synch sectors § Configuration and switch-over enhancements – No longer communication with HMC – Faster switch for around 20 -30 sec per IOA § Enhanced detection of primary DS HMC failure - we will now detect a DSCLI code freeze on the primary HMC § Toolkit supports Flash. Copy SE © 2008 IBM Corporation

IBM System Storage® Example: Start Flash. Copy of IASP STRFLASH ENV(<name of IASP>) Steps performed: ENDFLASH • Check that the server jobs are active Steps performed: • Check that clustering is correctly configured § Vary off all the Flash. Copy IASPs • Autostart any nodes in the cluster that are required § Modify clustering to enable a clean up. § IPL the backup system to clean up its virtual addresses. • Send an information message to operatior on the Production system • Vary off the IASP on production system • Run the mkflash script using the DSCLI • Vary-on the production copy of the IASP • Release/Reset the IOP/IOA resources on the backup partition • Make the IASP available to the backup system © 2008 IBM Corporation

IBM System Storage® Example: Switch Metro Mirror Steps performed: § Send *INQ message to QSYSOPR on the Production node § Vary off the IASP on the current Production node. § Run Metro Mirror Failover tasks for the IASP device using DSCLI. § Release/Reset the IOP/IOA resources on the backup partition and get the disks to register correctly as the IASP. § Vary on the IASP on the current Backup node (which then becomes the Production node). © 2008 IBM Corporation

IBM System Storage® Example: Display Metro Mirror Environment © 2008 IBM Corporation

IBM System Storage® IBM Power. HA for i § § § Product Number – 5761 -HAS Supported Release – i 5/OS V 6 R 1 Prerequisites – HA Switchable Resources (option 41) The IBM i 5/OS HA/DR clustering solution – Simple web-based management – Provides Cross Site Mirroring (XSM) function via • i 5/OS Geographic Mirroring • DS 8000 Metro Mirror • DS 8000 Global Mirror © 2008 IBM Corporation

IBM System Storage® Power. HA Structure Cluster mgmt, User inetrfaces LP 5761 -HAS Solution Based Task Based Commands XSM SAN Solutions Switchable resources Cluster technology DSCLI Option 41 - HASR i 5 OS Metro Mirror Geographic Mirroring Global Mirror Switchable Devices & Disk Flash Copy Source/Target Side Tracking IASP infrastructure CRG infrastructure Admin Domain infrastructure © 2008 IBM Corporation

IBM System Storage® Power. HA with DS 8000 Copy Services System i Cluster, Device domain DS/ ESS Online DS CLI IASP Power. HA Metro Mirror DS CLI Power. HA IASP IOA § Flash. Copy and Metro Mirror of IASP, Global Mirror will follow § GUI for creating cluster and IASP § GUI for Flash. Copy and Metro Mirror of IASP © 2008 IBM Corporation

IBM System Storage® Metro Mirror Scenario Setup: § Define and connect LUNs on DS 8000 § Create IASP on Production system with DS 8000 LUNs § Create cluster § Create Device domain § Create Device CRG § Install DS CLI on System i § Create Metro Mirror session § Create Copy description for the IASP on primary and backup node § Start CRG © 2008 IBM Corporation

IBM System Storage® Metro Mirror scenario - continue Scenario for planned outages : • Make IASP unavailable on primary node • Detach Metro Mirror session on primary node at primary site • Stop device CRG • Change roles in recovery domain • Restart device CRG • Reattach Metro Mirror ASP session on backup node at primary site • Make IASP available on primary node at secondary site © 2008 IBM Corporation

IBM System Storage® Flash. Copy Scenario Setup: § Define and connect LUNs on DS 8000 § Create IASP on Production system with DS 8000 LUNs § Create cluster § Create Device domain § Install DS CLI on System i § Create Copy description for the IASP on primary and backup node © 2008 IBM Corporation

IBM System Storage® Flash. Copy Scenario - continue Scenario for daily backups: Day 1 § CHGASPACT suspend § STRASPSSN § CHGASPACP resume § Perform backup of application from Backup node § CHGASPSSN *DETACH © 2008 IBM Corporation

IBM System Storage® Flash. Copy Scenario - continue Every other day § CHGASPACT suspend § CHGASPSSN *REATTACH § CHGASPACT resume § Perform backup of application from Backup node § CHGASPSSN *DETACH © 2008 IBM Corporation

IBM System Storage® Creating Metro Mirror session © 2008 IBM Corporation

IBM System Storage® Creating Copy Description © 2008 IBM Corporation

IBM System Storage® Detach MM session © 2008 IBM Corporation

IBM System Storage® Introducing IBM Total. Storage Productivity Center Standard Edition Productivity Center For Data Productivity Center For Fabric Productivity Center For Disk Productivity Center For Replication Limited Edition © 2008 IBM Corporation

IBM System Storage® What is TPC for Replication? § Volume level Copy Service Management § Coordinates Copy Service Functionalities – Flash Copy – Metro Mirror – Global Mirror – Metro Global Mirror § Ease of Use – Single common point of control – Web browser based GUI – Wizard based configuration © 2008 IBM Corporation

IBM System Storage® Simple Setup 1. Add Storage Device 2. Setup Paths via Wizard 3. Create Session 4. Add Copy Sets –Box to Box –LSS to LSS –Volume to Volume –CSV Import NEW 5. Issue Start! NEW © 2008 IBM Corporation

IBM System Storage® TPC for Replication with System i § Flash. Copy, Metro Mirror,

IBM System Storage® TPC for Replication – Global Mirror Recovery © 2008 IBM Corporation

IBM System Storage® Managing Tools for Solutions with External Storage Copy Services for System i (Toolkit) HASM TPC-R Flash. Copy of IASP Yes No Full System Flash. Copy Yes No Storage part Metro Mirror of IASP Yes No Full System Metro Yes Mirror No Storage part Global Mirror of IASP Yes No Fully System Global Mirror Yes No Storage part © 2008 IBM Corporation

IBM System Storage® How Does it Work? - i. Cluster Architecture Primary Node/LPAR Backup Node/LPAR i. C Read ERP TCP/IP Memory Cache Remote Journals i. C Apply DB 2 Match Merge DB 2 • Remote journal based • High speed send apply • Configure once on source © 2008 IBM Corporation

IBM System Storage® i. Cluster Switching Primary Node/LPAR Backup Node/LPAR i. C Read ERP Memory Cache High Speed Memory Cache TCP/IP Journals Match Merge i. C Read Remote Journal Remote Journals Journal i. C Apply i. C Apply DB 2 Match Merge DB 2 • User exits before and after switch for automation • Journaling automatically started on new source • Typical switch time in minutes © 2008 IBM Corporation

IBM System Storage® Geographic Mirroring © 2008 IBM Corporation

IBM System Storage® Overview of Geographic Mirroring § Storage management based page level replication. § Geographic mirroring runs over TCP/IP § Up to four communication interfaces on multiple adapters can be configured § Disk – Can be internal or external – Type, size and number of disk units may be different § Detach / Reattach © 2008 IBM Corporation

IBM System Storage® IBM i 5/OS Multi System Data Resiliency Switcheable IASP Clusters Switched Disk Cluster § Switched Disk § Strategic building block for IBM i 5/OS High Availability and Disaster Recovery solutions § Storage agnostic § Utilizes i 5/OS® Cluster Resources for automation and application interface § Single copy of data § HA (Local only) Geographic Mirroring Cluster Geographic Mirroring Metro Mirror/Global Mirror Cluster § Geographic Mirroring § Storage agnostic § HA and DR Flash. Copy Metro Mirror § Global Mirror § Flash Copy § HA and DR Cross Site Mirroring Logical Replication i. Cluster § Storage agnostic § HA and DR § Concurrent access § Supports i 5/OS Cluster Basic San Copy Services § Global Mirror § Boot From SAN § DR/Tape Backup Flash. Copy Global Mirror © 2008 IBM Corporation

IBM System Storage® Comparing Different Business Continuity Solutions i. Cluster Geo Mirroring Metro Mirror of IASP Global Mirror of IASP Metro Mirror full space Gloabl Mirror full space Recovery time Measured in min Less than 30 min (? ) Less than 30 min IPL Recovery point If remote journaling, zero Zero (? ) If journaling, zero Depends on GM interval Production system resources CPU , disk memmory usage CPU usage - - Usage of recovrey system Can be used for other workload Sysbas can be used for other workload Must be standby Performance impact Some impact Minimal Near zero Some impact at certain workloads Near zero Resynchronizat ion Incremental Incremental Administartion and maintenance Significant Need to miantain *sysbas - - Automation Autmated Can be done Automated Support Ca be concern in some countires ? Fully Supported Fully Supported Fully supported © 2008 IBM Corporation

IBM System Storage® Positioning: Questions to consider § How big is System i workload ? § How much can my data be behind at recovery ? § How quickly must I be up and runnung ? § Do I need DR solution for System i and other servers ? § Do I need DR solution for multiple System i ? § Which distance is suitable for my DR ? § What bandwidth can I afford ? § How reliable are the links ? § Do I need DR system for testing/developing ? § Is my system heavily used by my applications ? © 2008 IBM Corporation

IBM System Storage® Feature Storage Based Benefit ü Replication of all data written to disk IBM Metro Mirror ü Virtually unlimited geographic distance when using asynchronous IBM Global Mirror ü Generally implemented for DR* and HA* (synchronous) IBM Flash. Copy ü Can be cluster controlled ü Synchronous (Metro Mirror only) or asynchronous network transmission (Global Mirror only) ü No manual management of replicated objects required ü Flash. Copy means second copy of data can be used for fast backups, development, testing, etc. ü Order of changes is preserved via DS 8000 and Metro Mirror ü Resolution of out of synch conditions (objects or data) is under control of DS 8000 Operating System Based Products ü Page replication controlled by operating system ü Ordering of changes is preserved by the operating system ü No manual management of replicated objects required ü Synchronous network data transmission ü Is cluster controlled ü Supports external or internal storage Can be used alone or with IASPs & Toolkit Note: used with IBM Total. Storage DS 8000 Independent Auxiliary Storage Pools (IASP) Cross Site Mirroring (XSM) High Availability Solutions Manager (HASM) (implemented with XSM and IASP) (V 6 R 1 only) Considerations ¨ Network Bandwidth (all data and objects are replicated to second system) ¨ Only supported via external SAN storage ¨ No concurrent access to second copy of data ¨ IF not using IASPs, system IPL required for switching to backup copy ¨ Additional disk for Consistency Groups (Global Mirror only) ¨ Network Bandwidth (all data and objects in an IASP are replicated to the second system) ¨ Distance limitations ¨ No concurrent access to second copy of data ¨ Prior to V 5 R 4 no source side ü Only vary on of IASP is required for switching to backup system tracking ü Resolution of out of synch conditions (objects or data) is under control of i 5/OS ¨ Requires IASP implementation ü Generally implemented for HA* Logical Replication ü Selective data/object replication may reduce network overhead IBM: ü Virtually unlimited geographic distance IBM Data. Mirror i. Cluster. TM ü Concurrent access to second data copy (for backups, read only access, etc. ) Maximum Availability: ü No special hardware configuration requirements ü Choice of synchronous or asynchronous network transmission of data (synchronous only with remote journaling) www. maximumavailability. co m Trader’s: ü Generally implemented for DR* and HA* http: //www. quickedd. com/uk/index_uk. htm ü Can be cluster controlled Vision Solutions: ü Can be used alone or with other HA technology choices http: //www. visionsolutions. c om ü Supports external or internal storage ü No IPLs required for switching. When used in conjunction with IASP— vary on of IASP is required ü Ordering of changes controlled by HA ISV software interfaces ¨ Possible apply lags on target system for changes in high volume environments ¨ Management to ensure all new objects are replicated ¨ Bandwidth has to be sufficient for replicated objects and data , and to ensure no data send lag ¨ Resolution of out of synch conditions (data or objects) is not under control of i 5/OS © 2008 IBM Corporation