40
] Orange County Convention Center Orlando, Florida | May 15-18, 2011 Near Zero Downtime @ Nike Courtney Hofmann – Nike Marc Hopkins – Nike Mariusz Debowski – SAP

3501 How Nike Minimized System Downtime With the Near Zero Downtime Method Through ALM

Embed Size (px)

DESCRIPTION

501 How Nike Minimized System Downtime With the Near Zero Downtime Method Through ALM

Citation preview

  • ]Orange County Convention CenterOrlando, Florida | May 15-18, 2011

    Near Zero Downtime@ Nike

    Courtney Hofmann NikeMarc Hopkins Nike

    Mariusz Debowski SAP

  • Real Experience. Real Advantage.

    [

    2

    Learning Points

    ? Understand how Nike was able to leverage Near Zero DownTime (NZDT) to reduce the Business Downtime for itsUpgrade and Unicode Conversion project.

    ? Understand the key differences between the NZDTmethodology and a standard Upgrade & Unicode Conversionapproach.

    ? Understand the challenges & benefits introduced by the NearZero Down Time methodology.

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    3

  • Real Experience. Real Advantage.

    [ Costs of downtime vs. cost of availability

    ? Downtime costs are usually not linear over thetime of system outage

    ? They depend on the business impact? For longer downtimes costs can increase

    progressively? Example: when SCM processes are stuck longer

    than 3h the whole production will be stuck andcause even higher costs

    Availability Cost Curve

    Cos

    t

    95% 98% 99,5% 99,9%

    Availability

    (Source: Gartner Group)

    Costs

    ? Redundant components? Disaster recovery site? System management tools? IT staff? Capacity planning? Guaranteed SLA? Proactive services

    downtime

    cost

    s

    4

  • Real Experience. Real Advantage.

    [ Business Continuity RoadmapCurrent Areas of Investment

    SPs EhPs Upgrade

    Any system

    NW hub

    App specific

    SAPehpi (for EHPs &SP Stacks only)

    SAPup

    nZDMEP/PI

    NZDT (ERP)

    Standardtools

    Customerprojects

    XXL Customers

    EnBW: ~30min Business downtime

    SBSA: ~ few hoursEhP5 TCO lab:3h tool downtime,10h tool runtime

    New Iterative Shadow Upgradeready for pilot end of 2011

    Extension to Banking Servicescurrently in evaluation

    Further optimizationand improvements

    5

  • Real Experience. Real Advantage.

    [ Downtime Minimization

    X00s customersrunning global systems in7x24x365 mode

    X000s customersrunning mission criticalapplications

    X0000s customerswith less demand on businesscontinuity

    Need forBusinessContinuity

    Need forminimalOutages

    Need forlowest TCO &simplicity

    Prio 1 Demand

    Near Zero Downtime /Iterative Shadow Upg

    SAP ServicesBusiness Downtime Optimization

    SAP tools easy to use andappropriate documentation

    SAP is constantly working on supporting our customers in the effort tominimize their system downtimes

    Further optimization of the existing tools for software updates

    Sharing with the customers the collected experience and best practices

    Creation of new tools and service offerings for customers operating their systems in 24x7 mode

    6

  • Real Experience. Real Advantage.

    [ 10 steps to Near Zero DowntimeHigh level illustration

    Delta replay

    Recording

    PRDR/3 4.6C nUC

    hostA

    clone

    downtime

    Prepare + Upgrade Uptime

    Upgradeand Unicode conversion

    + Transports+ Add Ons+ Pre-Testing

    Last Delta replay

    Final-Validation

    Infrastructureadjust. -interfaces

    Upgrade + UC Downtime Post ProcessingDeltaTransfer

    Infrastructure

    PRDR/3 4.6C nUC

    hostB

    PRDECC 6.03 UC

    hostB

    hostA

    Validation

    2

    3

    45

    6

    7 8

    9

    10

    21

    1. Users are working on PRD system2. Copy of PRD system & start recording of transaction data3. Data from Production live transactions are recorded4. Upgrade and Unicode conversion on the Copy to ERP 6.035. Finish EhP implementation and replay transactions from Production to Copy6. Users shutdown on Production system and finish replaying transactions7. Start reconciliation source to target system8. Get signoff that both systems are reconciled9. GO decision10. Start up users on Copy (new PRD)

    DO

    WN

    TIM

    E

    7

  • Real Experience. Real Advantage.

    [ Near Zero DowntimeApplicability use cases

    The method can be used for numerous downtime events? Release Upgrade

    ?Enhancement Package implementation?Support Package Stack installation?Unicode conversion?Customers release mass transports

    and non-SAP events?OS/DB migration?OS patches?DB patches?DB reorgs

    and any combination of events above

    Delta replay

    Recording

    PRDERP 6.02

    hostAclone

    downtime

    Prepare + Upgrade Uptime

    Implement EhP 4 (and others)

    + Transports+ Add Ons+ Pre-Testing

    Last Delta replay

    Final-Validation

    Infrastructureadjust. -interfaces

    EhP installation downtime Post ProcessingDeltaTransfer

    Infrastructure

    PRDERP 6.02

    hostB

    PRDERP 6.04

    hostB

    hostA

    Validation

    ? 8

  • Real Experience. Real Advantage.

    [

    Near Zero Downtime @ Nike

    1st combined Upgrade & Unicode Conversion using NZDT

    9

  • Real Experience. Real Advantage.

    [ The Upgrade A New SAP Landscape

    10

    Opportunity for Dramatic Technical Changes

    ? SAP Upgrade from 4.6C to ERP 6.0? New Oracle Version - 10.2.0.4? New Hardware HP Superdome Virtual Partitions with shared CPUs? New Storage Frame EMC VMax? New Network Infrastructure? Unicode conversion performs a full database reorg of all tables & indexes? Oracle Index Key Compression? Complete Technical RE-DO of existing system

  • Real Experience. Real Advantage.

    [

    ? Three parallel upgrades/Unicode Conversions were in scope (25TB, 12TB, ~1 TB)? First Upgrade & Conversion took 16 days (18TB)? The challenge was to reduce it to 7 days? Reduced the standard upgrade & Unicode conversion to 4 days? System continued to grow at 500GB / month causing downtime to increase (25TB)? 4+ days was still too long for the business

    Why Nike Chose NZDT

    How much downtime can we afford?

    11

    Building Blocks of the Standard ERP Upgrade

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    12

  • Real Experience. Real Advantage.

    [ Nikes ERP System Landscape

    PRA320

    PRD300

    Config. Development

    PTE300

    PTA320

    TransportPath

    110110 120120

    168168

    ER1

    100100

    Security

    130130

    PTF330

    PRF330

    0.5TB9 TB21 TB

    Single shared code &configuration base

    Single developmentenvironment is source of alltransports

    All 3 systems follow the sameupgrade path

    13

    Landscape Details

  • Real Experience. Real Advantage.

    [ Nike ERP Upgrade - Release ScopeBeforeBefore

    SAP R3 4.6c 10 year old system

    16 languages implemented via MDMPTechnology

    Old hardware nearing end of life

    1695 Modifications 182 Clones 13 Special Developments 459 SPDD / 2521 SPAU Historically grown enhancements

    Language specific Interfaces

    Twin Upgrade and Unicode ConversionTarget

    6+ days of Business Downtime

    AfterAfter

    SAP ERP 6.0

    Unicode Fully reorganized database Standardized & optimized database layout

    New Hardware/Infrastructure Footprint

    870 Modifications 81 Clones 7 Special Developments 87 SPDD / 1395 SPAU Consolidated redesigned enhancements

    in key areas

    XML based UTF8 Interfaces in SAP

    Near Zero Down Time

    Weekend Outage window

    BasisPlatform

    NikeCustom

    Solutions

    UpgradeProcess

    14

  • Real Experience. Real Advantage.

    [ Standard Upgrade vs. Near Zero DowntimeStandard UpgradeStandard Upgrade

    Major upgrade phases executed duringdowntime? PRD: 69 hours

    Unicode Conversion executed duringProduction System downtime? PRD: 96 hours

    4000 Transports imported duringdowntime? PRD: 20 hours

    NZDTNZDT

    Major Upgrade changes imported whileProduction systems were available

    Unicode Conversion executed whileProduction systems were available

    4000 Transports imported while Productionsystems were available

    Upgrade

    UnicodeConv.

    PostUpgrade

    Tasks

    15

  • Real Experience. Real Advantage.

    [ Nikes ERP Upgrade Project History

    2011201020092008

    InitialUpgrade

    Effort(Dev & QA)

    UpgradeProof ofConcept Upgrade

    DevelopmentTUUC*

    Methodology

    NZDTProof ofConcept

    NZDTProject

    UpgradeQA Systems

    GoLive!

    *TUUC: Twin Upgrade & Unicode Conversion

    16

  • Real Experience. Real Advantage.

    [ Nike ERP Upgrade High-Level Project Requirements

    ? Minimum impact to productive operations due to upgrade

    ? Business Downtime as low as possible

    ? Allow flexibility to reschedule upgrade within a short timeframe

    ? One-step Upgrade procedure

    ? In case of problems the upgrade must be completely reset-able(also possible using snapshot technology established at Nike)

    Nik

    e R

    equi

    rem

    ents

    SAPs Near-Zero-Downtime Technology

    Next StepNext Step

    17

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    18

  • Real Experience. Real Advantage.

    [ Nike NZDT Project Schedule20102009

    ? Code Remediation? QAST System Test? SIT Integration Test? UAT User Acceptance Test? Dry Run? Go-Live

    ? NZDT POC Proof of Concept? NZDT SAP Analysis? QAST Build (NZDT without delta)? NZDT Full execution 1? NZDT Full execution 2? NZDT Dry-Run Build

    Legend: Duration Milestone

    2011

    J JA O JF M M J A S N DDNO F M

    Upgrade Go-Live with 3parallelsystems.

    Total number of builds is 15 (5build cycles with 3 systems each).4 build cycles need completeNZDT procedure.

    19

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    20

  • Real Experience. Real Advantage.

    [ Stepping Through Nikes Upgrade with NZDT

    21

  • Real Experience. Real Advantage.

    [ Cutover Timeline Uptime & Downtime

    13-D

    ec

    14-D

    ec

    15-D

    ec

    16-D

    ec

    17-D

    ec

    18-D

    ec

    19-D

    ec

    20-D

    ec

    21-D

    ec

    22-D

    ec

    23-D

    ec

    24-D

    ec

    25-D

    ec

    26-D

    ec

    27-D

    ec

    28-D

    ec

    29-D

    ec

    30-D

    ec

    31-D

    ec

    1-Ja

    n

    2-Ja

    n

    PREPARE, UpgradeUptime Clone NZDT Recording Window Down-time

    CloneUpgrade

    Down-time,SPUMG from UAT

    UnicodeConversion

    UnicodeConversion

    Tran

    spor

    ts

    Dat

    a C

    onve

    rsio

    ns

    SUM

    G

    Del

    ta R

    epla

    y

    Dat

    a C

    onve

    rsio

    ns

    Test

    ing

    ProductionR/3 4.6C

    NZDT CloneUC Source

    UC TargetFuture ERP 6.0

    22

  • Real Experience. Real Advantage.

    [ Cutover Timeline The Downtime Window12/31 1/1 1/21/3

    RampDown

    Final Replay, HW Reconfiguration, Final Build Steps,Upgrade Conversions, Manual Config, FI Audit Snapshots

    FI Ramp Up,Link Testing

    Control Test

    Timeline Callouts

    ? Ramp Down includes system lockout, FI Audit Snapshots, Delta Queue Cleanouts.? Final Delta Replay: Good performance and no unexplainable differences? Upgrade Conversions: Good performance? Total downtime duration 79 hrs

    (from system lockout until Go Decision)

    ? Hardware reconfiguration required 8 hours? Not using NZDT would have resulted in at least

    an extra 6 days of downtime!

    23

  • Real Experience. Real Advantage.

    [ NZDT Validation

    SAP2007 /

    ? Technical Reconciliation? Initial plan:

    Highly complex automated comparisonbetween NZDT Upgrade andnon-NZDT Upgrade system (Tech Rec 1)

    ? Summer / Fall 2010:SAP developed direct 4.6c to ERP 6.0reconciliation covering (Tech Rec 2)? Release changes? Unicode ConversionVia Table & Field level comparisons? Enabled ability to validate build during Cutover downtime

    ? Business Testing? Unicode Conversion? Unix Scripts for row-counts

    0

    50

    100

    150

    200

    250

    300

    350

    Time in hrs

    Tech Rec 1 Tech Rec 2

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    25

  • Real Experience. Real Advantage.

    [ Challenges Project

    Issue Description Resolution

    NZDT Performance Impactto PRD

    NZDT recording puts additional load onproduction systems

    Multiple recording tests in Production

    NZDT Recording FreezeTriggers

    Frozen tables cause dumps in Production 24 x 7 monitoring by SAP & NikeDaily review meeting

    Large amount of extrahardware

    Multiple copies required of each Productionsystem

    Virtualization and Cloning technique

    Highly complex process Recording, mapping and replay technologycreates a highly complex process

    Partnership with SAP enabled accessto expert resources

    Security Access Unique NZDT process did not allow for theuse of Fire Fighter tool

    Limited number of core teammembers with build access

    Language Remediation forUnicode Conversion

    On-going difficulties to establish ownershipof language remediation tasks

    Small expert team performedremediation during Cutover

    26

  • Real Experience. Real Advantage.

    [ Agenda1. Near Zero Down Time Overview

    2. Nike Upgrade Project - Overview

    3. NZDT Project Schedule

    4. NZDT Project Execution

    5. NZDT Challenges

    6. NZDT Learning's

    27

  • Real Experience. Real Advantage.

    [ Nike Learnings Overall project

    ? Small cohesive core team with technical expertise used iterative execution& refinements for complete NZDT solution.

    ? Excellent Cooperation between Nike and SAP teams

    ? Perseverance of entire team through extreme hours and complexity toachieve objectives.

    ? Co-location of project team kept everyone on the same page regardingstrategy, current activities, planned activities, issues, and status

    28

  • Real Experience. Real Advantage.

    [ Nike Learnings Architecture

    ? Virtualization and Cloning technology set the stage to enable futureflexibility, rapid provisioning of ERP QA systems

    ? Initiate discussions with infrastructure partners early in the process

    ? Use of Oracle Flashback to instantly restore to a specified point in timewas vital (was utilized in every build)

    ? Plan for enough hardware to allow flexibility and ability to adapt tochanges in requirements

    29

  • Real Experience. Real Advantage.

    [

    ]? Thank you for participating.SESSION CODE:3501

    Please remember to complete and return yourevaluation form following this session.

    For ongoing education in this area of focus, visit www.asug.com.

  • Real Experience. Real Advantage.

    [ Appendix

    31

  • Real Experience. Real Advantage.

    [ Nike ERP Landscape before Upgrade

    PRA320

    PRD300

    Config. Development

    PSQ300

    PSA320

    ProdSupport

    110 120

    168

    DE1

    100

    Security

    Prod Support

    130

    PSF330

    PRF330

    ER1

    100

    PTA320

    110 120

    130

    PTE300

    PTF330

    12/10Upgrade

    168

    SecurityConfig. Development

    32

  • Real Experience. Real Advantage.

    [ Upgrade Prepare PhasePRA

    320PRD

    300

    110 120

    168

    DE1

    100

    Prod Support

    130

    PRF330

    ER1

    100

    PTA320

    110 120

    130

    PTE300

    PTF330

    12/10Upgrade

    168

    SecurityConfig. Development

    33

  • Real Experience. Real Advantage.

    [ Delta Recording Upgrade & Unicode Conv.PRA

    320PRD

    300

    PRD300

    PRA320

    110 120

    168

    DE1

    100

    Prod Support

    130

    PRF330

    PRF330

    ER1

    100

    PTA320

    110 120

    130

    PTE300

    PTF330

    12/10Upgrade

    168

    SecurityConfig. Development

    Upgrade & Unicode Conversion

    DeltaRecording

    34

  • Real Experience. Real Advantage.

    [ Delta ReplayPRA

    320PRD

    300

    PRD300

    PRA320

    110 120

    168

    DE1

    100

    Prod Support

    130

    PRF330

    PRF330

    ER1

    100

    PTA320

    110 120

    130

    PTE300

    PTF330

    12/10Upgrade

    168

    SecurityConfig. Development

    DeltaRecording Delta

    Replay

    35

  • Real Experience. Real Advantage.

    [ Go Live

    PRD300

    PRA320

    PRF330

    ER1

    100

    PTA320

    110 120

    130

    PTE300

    PTF330

    12/10Upgrade

    168

    SecurityConfig. Development

    36

  • Real Experience. Real Advantage.

    [ Near Zero DowntimeCustomers cases

    Customers successful with Near Zero Downtime

    Airbus EMEA, IS-Aerospace & Defense, Aircrafts Spare Parts ManagementRelease upgrade from SAP R/3 4.6C to SAP ERP 6.0First successful implementation of the NZDT method in June 2009System size 2 TBDatabase OracleAchieved technical downtime 4 hrsAchieved business downtime 6 hrs

    Nestl EMEA, Nutrition and Food IndustryRelease upgrade from SAP R/3 4.7 to SAP ERP 6.03Successful GoLives of 9 ERP systems with the NZDT method in Spring 2010Total system size 80 TB (all 9 systems)Achieved technical downtime 2 hrsAchieved business downtime 16-20 hrs

    Customer B EMEA, Diary ProductsRelease upgrade from SAP R/3 4.7 to SAP ERP 6.03 combined with Unicode conversionSuccessful GoLive of the ERP system with NZDT in October 2010Total system size 2 TBDatabase OracleTechnical downtime 3:30 HrsBusiness downtime 12 hrs

    ? 37

  • Real Experience. Real Advantage.

    [ Near Zero DowntimeCustomers cases

    Customers with running Near Zero Downtime projects

    Nike USA, IS-Foot & ApparelRelease upgrade from SAP R/3 4.6C to SAP ERP 6.03 combined with Unicode conversionSuccessful GoLive of 3 ERP systems with NZDT in January 2011Total system size 40 TBDatabase OracleTechnical downtime 18 HrsBusiness downtime 79 hrs

    Customer D EMEA, Automobile and Industrial Equipment and Consumer GoodsCustomers release including implementation of EhP4Successful GoLive of 5 ERP system with NZDT in November 2010NZDT will be used on a regular basisTechnical downtime 3 HrsBusiness downtime 20 Hrs

    Customer E USA, High-Tech IndustryRelease upgrade from SAP R/3 4.7 to SAP ERP 6.03 combined with Unicode conversionPlanned GoLive of the ERP system with NZDT in August 2011Database OracleTechnical downtime ? HrsBusiness downtime 15 hrs

    ? 38

  • Real Experience. Real Advantage.

    [ Near Zero DowntimeCustomers cases

    Customers with running Near Zero Downtime projects

    Customer F EMEA, Consumer ProductsRelease upgrade from SAP R/3 4.7 to SAP ERP 6.04 combined with Unicode conversionSLO projectSuccessful GoLive of the ERP system with NZDT in February 2011Database OracleTechnical downtime 19 HrsBusiness downtime 48 hrs

    Customer G EMEA, Household ProductsCustomers release including implementation of EhP4Successful GoLive of ERP systems with NZDT in January 2011NZDT will be used on a regular basisTechnical downtime 4 HrsBusiness downtime 20 Hrs

    ? 39

  • Real Experience. Real Advantage.

    [

    ]? Thank you for participating.SESSION CODE:3501

    Please remember to complete and return yourevaluation form following this session.

    For ongoing education in this area of focus, visit www.asug.com.