1 / 42

Technology Spotlight on EMC Symmetrix Remote Data Facility

Technology Spotlight on EMC Symmetrix Remote Data Facility. Agenda. Enterprise Storage SRDF Introduction Business Challenges SRDF Configurations SRDF Operational Modes SRDF Operation. ENTERPRISE STORAGE. Time Finder TM. Enterprise Storage Framework. Information

melva
Download Presentation

Technology Spotlight on EMC Symmetrix Remote Data Facility

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Technology Spotlight on EMC Symmetrix Remote Data Facility

  2. Agenda • Enterprise Storage • SRDF Introduction • Business Challenges • SRDF Configurations • SRDF Operational Modes • SRDF Operation

  3. ENTERPRISE STORAGE Time FinderTM Enterprise Storage Framework Information Sharing Information Protection Information Management Information Protection Information Management Information Sharing ESP SMTF Data ReachTM SNFS Time FinderTM SDMS Symmetrix Manager FDRSOS EDM Internet Services ESP SMTF Data ReachTM SNFS Time FinderTM SDMS Symmetrix Manager FDRSOS EDM Internet Services SRDF

  4. SRDF Introduction

  5. What Does SRDF Do? • SRDF creates and maintains real time or near real time physically separate copy of data, without using host cycles, in a local or remote EMC Symmetrix that can be directly addressed by a designated Host system even in the event of individual drive or link failures.

  6. Local Volume May be spared, locally mirrored, unmirrored or part of a RAID-S group Source Volume An active application volume containing production data and programs. The Standard Volume may be locally mirrored, unmirrored, part of a RAID-S group, and/or covered by a global spare. Target Volume A volume that contains the copy of a Source Volume Synchronized Volumes The background task performed by Symmetrix to create an identical mirror image of two volumes; as in RAID 1One volume is always Synchronized to another. RLD, Remote Link Director Symmetrix SRDF Communication feature Configuration Types Unidirectional Bi-directional Uni-directional Primary Operational Modes Synchronous Semi-synchronous Secondary Operational Modes Adaptive Copy Domino Invalid Tracks Other Definitions M1: A Source Volume M2: The RAID 1 mirror of the M1 R1: Source side of Remote Mirror R2: Target side of Remote Mirror SRDF Nomenclature

  7. Symmetrix Remote Data Facility • Logically synchronized versions of selected volumes • Independent of CPU, operating system, application or database • Resilient against drive, link and server failures • Selectable synchronization behavior Campus (66Km) Extended Distance (T1/T3, E1/E3, ATM)

  8. Business Challenges

  9. The Business Challenge • The Business Problem • Data inaccessibility measured in $$$$ • Continuous availability for all data becoming the norm • Disaster recovery for business receiving renewed focus • Symmetrix Remote Data Facility • Maintain real-time or near-real-time physically separate copy of selected volumes • Use no host CPU resources • Operating system independent • Continue running through events such as individual drive/link failures

  10. SRDF Applications • Business Continuance • Act as separate mirror for business and application continuance • Disaster Recovery • Very rapid data availability with data current to last write • Data Center Migration • Effect data center move with minimum application outage • Workload Management • Provide service through scheduled outages such as maintenance

  11. Disaster Recovery with SRDF • Disaster recovery across the enterprise • Consolidates heterogeneous platforms into one DR solution • Restart applications at recovery site with no data restore required • Maximized application availability • Test Disaster Recovery process in place • Validate completeness of DR plan to minimize “surprises” • Return home after disaster recovery with minor disruption • Synch and re-synch automated and transparent to users

  12. SRDF Configurations

  13. ... ... Local Source Local Target Active Path Recovery Path SRDF Uni-Directional Configuration Optional CPU(s) CPU(s) RLD 1 RLD 1 RLD 2 RLD 2 Fibre, ATM, or T1/T3 SITE A SITE B

  14. ... ... Local Source Local Target Target Source Active Path Recovery Path SRDF Bi-Directional Configuration CPU(s) CPU(s) RLD 1 RLD 1 RLD 2 RLD 2 Fibre Only SITE A SITE B

  15. ... ... Local Source Local Target Target Source Active Path Recovery Path Dual SRDF Configuration CPU(s) CPU(s) RLD 1 RLD 1 RLD 1 RLD 1 RLD 2 RLD 2 RLD 2 RLD 2 Fibre, ATM, or T1/T3 SITE A SITE B

  16. Campus Solution Private Fiber or Common Carrier 9036 9032 9033 9191 9036 9032 9033 9191 SITE A SITE B Active Channel Private or Common Carrier - 46-66 km max, Repeater required every 20-30 km Recovery Channel

  17. Extended Distance Solution ATM, T3 or E3 Other Network Device Other Network Device SITE A SITE B Active Channel Leased Carrier - Distance Limited to Carrier Recovery Channel

  18. Extended Distance: The Problem • Speed of light (186,000 miles/sec) is Constant! • 1ms delay per 125 miles ( x 2 for response) • 8ms for send and confirm across a 500 mile trek • Single Queue, Multiple Servers (RLDs). • For a 500 mile link - I/O from 6 devices queued • first 2 devices delayed 8ms each • next 2 delayed ~ 16ms each • last 2 delayed ~ 24ms each • links are idle 87% of the time - more for longer distances

  19. Dev 0 Dev 1 Dev 2 Dev 3 Dev 4 Extended Distance: Standard and FarPoint Queue DATA & Cmd • Multiple Logical Devices per link. • Device waits for confirming Response. RLD RLD STATUS SRDF Standard Mode • One block per link. • Link waits for confirming response. Queue RLD RLD Dev 0 Dev 1 Dev 2 SRDF FarPoint Dev 3 Dev 4

  20. Extended Distance: FarPoint • Send Multiple -- Confirm Multiple • Multiple devices, multiple servers (RLDs) • Queued in order received • Up to n transactions in the “Pipe” at any time • Variable Based on Bandwidth, Distance, and Block size

  21. FarPoint: Impact • Higher Level of Throughput • Leveling of Response Time • Base Response Times Not Changed • Speed of Light Unaffected. • More Impact at Higher Distances and Heavier Loads

  22. FarPoint: Special Considerations • Requires 5x64 microcode • Default for Extended Distance (T3/E3) • Supported • Synchronous mode • Semi Synchronous Mode • Adaptive Copy Mode • Not Supported • Bi-Directional Mode • Error Recovery • Flush the Pipe • Revert to Standard Mode

  23. SRDF Operational Modes

  24. SRDF Modes of Operation • Synchronous • Semi-Synchronous • Adaptive Copy • Disk Mode • Write Pending Mode SRDF Processes I/O to Remote Symmetrix in Serialized Queues SRDF Updates I/O to Remote Symmetrix Periodically Based on Time Slice Setting

  25. Volume 16 Volume 21 Volume 5 Volume 62 Volume 3 Volume 10 SRDF Serialized Queue To RLD’s • Queue of Updates from Source Symmetrix to Remote Symmetrix • FIFO Queue : First-In-First-Out • Maximum of one I/O per Volume in the Queue at any Time From SCSI Adapters, ESCON Adapters, Fibre Adapters or Channel Adapters

  26. Local Source Local Target SRDF Synchronous Mode 1. Local Symmetrix receives a write from the Host 3. Receipt acknowledged by Remote Symmetrix 4. Ending status presented to Host 2. Local Symmetrix transmits data to the Remote Symmetrix

  27. SRDF Synchronous Mode Source Target Source Data Always Equals Target Data • Data integrity of all copies is maintained • Local and Remote Symmetrix maintain a synchronized copy of data • I/O is Serialized and placed on Symmetrix System Wide FIFO Queue • “I/O Elongation” - Duration of time for each I/O is increased due to serial activity PRIORITIES Data Integrity Disaster Recovery

  28. Local Source Local Target SRDF Semi-Synchronous Mode 1. Local Symmetrix receives a write from the Host 4. Receipt acknowledged by Remote Symmetrix 2. Ending status presented to Host 3. Local Symmetrix transmits data to the Remote Symmetrix

  29. ~ ~ SRDF Semi-Synchronous Mode Source Target Source Data Almost Equals Target Data Out of Synch by 1 I/O per Volume at most • I/O to Remote Symmetrix is done in parallel • Reads are executed even while writes are in transit to Remote Symmetrix • Additional writes wait until original write is acknowledged • Local and Remote Symmetrix maintain a nearly synchronized copy of data • I/O is Serialized and placed on Symmetrix System Wide FIFO Queue • “I/O Elongation” is eliminated PRIORITIES Performance Data Integrity Disaster Recovery

  30. Local Source Local Target SRDF Adaptive Copy Mode 1. Local Symmetrix receives writes from the Host 2. Ending status presented to Host 5. Receipt acknowledged by Remote Symmetrix 4. Local Symmetrix transmits data flagged as updated based on Time Slice to the Remote Symmetrix 3. SRDF Flags the Record for later Update to Remote

  31. a SRDF Queue of Adaptive Copy Updates To RLD’s • Updates are Time Sliced into Queue • Default Setting is 1 in 10 • Every Tenth I/O is an Update to the Remote Symmetrix • Setting can be changed at any time Volume 16 Volume 5 Volume 19 Volume 3 Volume 10 Volume 13 Volume 14 Volume 7 Volume 8 Time Slice Update Volume 17 From SCSI Adapters, ESCON Adapters, Fibre Adapters or Channel Adapters

  32. SRDF Adaptive Copy Mode Source Target Source Data does not Equals Target Data Possibly Out of Synch by ‘N’ tracks based on Symmetrix Activity • I/O to Remote Symmetrix is done in parallel • Reads and Writes are executed even while updates are pending or in transit to Remote Symmetrix • No Serialization of I/O data to Remote Symmetrix by Arrival Sequence • Host I/O placed on FIFO Queue periodically as determined by AD-COPY RATE • Data movement occurs with minimal impact on Local Host and Local Symmetrix • Adaptive Copy Configurations • Disk Mode • Write Pending Mode PRIORITIES Moving Data On Line Backup at Remote Site

  33. Local Source Local Target SRDF Domino • Domino Configuration Sets Volumes to NOT READY State if Any One of the Following Occurs: • Source Unavailable • Links Unavailable • Target Unavailable • Links Only Option Sets Volumes to NOT READY Only if Links are Unavailable Links Source Target

  34. Local Source Local Target SRDF Invalid Tracks • Notification of Invalid Tracks Provided • Stops Target Volumes from Becoming Ready When Multiple Links Fail Links Source Target

  35. SRDF Operation

  36. Error Reporting and Diagnostics • Message reporting to Host/Server console • Auto Call on abnormal conditions in HDA, link, subsystem • Remote support facility diagnostics

  37. Local Source Local Target Loss of Drive at Primary Site Primary Host Remote Host Links Primary Symmetrix Remote Symmetrix Source Target

  38. Loss of Drive at Site A • Application continues to run on Host/Server at Site A • Read/Write operation on affected drive go to Target volume at Site B • Exposure if Target volume lost (no local mirroring) • MESSAGE: with address of failed device • Repair drive at Site A • Command to resync: from Target to Source • MESSAGE: when resync. starts • MESSAGE: when resync. complete

  39. Local Source Local Target Loss of Primary Site Primary Host Remote Host Links Primary Symmetrix Remote Symmetrix Source Target

  40. Loss of Primary Site • Single command makes target drives accessible to recovery Host/Server CPU • Command can be issued from: • Service processor console • Remote support center • Local host using host component • Applications resumed within minutes at most recent status

  41. Runs as MVS subsystem Optional, chargeable, licensed per CPU Complements SRDF base system Initiated via started task Uses SAI to retrieve information from Symmetrix subsystem Provides status and configuration control Extends automated operations to contingency planning SRDF Host Component

  42. ENTERPRISE STORAGE MAINFRAME OPEN NETWORK DATABASE PLATFORM-SPECIFIC CONTROLLERS MEDIA Enterprise Storage Checklist Enterprise connectivity Cascadable Information-centric • Information protection • Information sharing • Information management Business Impact Operational Impact Financial Impact

More Related