Hear first-hand from Damon Anderson, Manager of Data Services at a leading Fortune 500 global distribution company, about how he researched and solved his IMS-to-DB2 migration challenge.
IMS applications continue to run many of today’s leading financial, manufacturing, and government agencies’ most critical business processes. However, we all know the developers that wrote and supported those applications are fast leaving the work force, creating a major skills gap, unwanted risk, and a pressing need for an efficient DB2 migration path.
While some organizations are having to turn to expensive 3rd party vendors to support these applications, the potential data security breaches and challenges in ensuring consistent quality of application support and maintenance make that a risky, expensive proposition.
Hear how Damon and his team mitigated these challenges by migrating their IMS application databases to DB2 with Syncsort DL/2 and how it helped them:
o avoid a skills gap
o eliminate their MLC costs for IMS DB and the significant costs for 3rd party IMS Support tools
o create a platform for future modernization of their core business applications
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
IMS to DB2 Migration: How a Fortune 500 Company Made the Move in Record Time With Minimal Disruption
1. Damon Anderson and Bill Bostridge
July 2016
IMS to DB2 Migration:
How a Fortune 500 Company Made the Move in
Record Time With Minimal Disruption.
2. Meet Today’s Presenters
2
Bill Bostridge
Director of Business Development
Syncsort
Damon Anderson
Manager of Data Services
Leading Fortune 500 Global Distribution Company
3. Syncsort is the Global Leader in Big Iron to Big Data Solutions
3
Syncsort Confidential and Proprietary - do not copy or distribute
• We are a leading data management software company and the
global leader in Big Iron to Big Data solutions for large enterprises
• Our innovative and high-performance software enables customers
to tap into the value of their data assets while dramatically reducing
the cost of managing mainframe and legacy systems
• Our Big Iron to Big Data solutions address the complex set of
requirements for delivering rich data from mainframe and legacy
sources to Big Data environments, powering business and
operational analytics
• A trusted industry leader for nearly 50 years, we deliver a unique
focus on customer value through cost-effective solutions and
unparalleled support
Marquee global customer base of leaders and
emerging businesses across all major industries
WOODCLIFF LAKE, NJ
JAPAN
SINGAPORE
Deep engineering and go-to-market relationships with
strategic partners in Big Iron and Big Data ecosystems
Marquee global customer base of leaders and
emerging businesses across all major industries
4. Syncsort DL/2 - Transparent Data Migration to DB2®
Migrate IMS Segments to DB2
tables without making any
application changes
Eliminate IMS DB and 3rd party
IMS support tools
Move application maintenance to
SQL and DB2
Syncsort DL2 is not a replication
tool, it is a replacement for IMS DB
Combination of software and
services for a low risk migration
from IMS to DB2
Syncsort Confidential and Proprietary - do not copy or distribute
4
5. What Do We Mean by Transparent Data Migration?
• No modifications, compiles, recompiles of existing application programs
• Migrate at your own pace – one database at a time or multiple databases
• Environmental changes for batch JCL and CICS can be made in non-
disruptively in advance
• Simple switch between IMS and DB2 for application testing
• Fallback capability
• Easiest and fastest route to DB2 and value delivery
LOWEST RISK MIGRATION APPROACH – PROVEN TIME AND TIME AGAIN
5
Syncsort Confidential and Proprietary - do not copy or distribute
6. DL/2: Transparent Migration Process Overview
6
Before During After
IMS
NO
DB2
YES
DL2 Stub
Application Program
Static SQL
Data
in
DB2?
IMS Stub
Application Program
DL2 Stub
Application Program
IMS
DB2
Syncsort Confidential and Proprietary - do not copy or distribute
7. DL/2: Transparent Migration Process Overview
7
Before During After
IMS
NO
DB2
YES
DL2 Stub
Application Program
Static SQL
Data
in
DB2?
IMS Stub
Application Program
DL2 Stub
Application Program
IMS
DB2
Syncsort Confidential and Proprietary - do not copy or distribute
8. DL/2: Transparent Migration Process Overview
8
Before During After
IMS
NO
DB2
YES
DL2 Stub
Application Program
Static SQL
Data
in
DB2?
IMS Stub
Application Program
DL2 Stub
Application Program
IMS
DB2
Syncsort Confidential and Proprietary - do not copy or distribute
9. Accessing IMS Databases
9
DBCTL
DLISAS DBRC
IMS/DC
CICS/TS
BMP
DLI/DBB
B
A
T
C
H
Supported by DL/2:
• xxxTDLI Assembler IMS database calls
• EXEC DLI COBOL IMS system calls
• AIBTDLI PL/I and LE/370 IMS command
codes
Syncsort Confidential and Proprietary - do not copy or distribute
11. 11
Why Migrate from IMS to DB2
Save Money
Cost of IMS Software
Cost of Software to monitor IMS
Cost of Software to Backup, Recover,
Reorganize, Validate
Personnel / Consultants
Staffing Shortages
Fewer and fewer developers and DBA’s
with IMS Skills
Consolidate to one z/OS DBMS
Simplify
12. 12
More Why Migrate off of IMS
Forced to upgrade IMS to stay on supported
release.
IMS 11 – GA 2009 – EOS 2014
IMS 12 – GA 2011 – EOS 2016
IMS 13 – GA 2013
IMS 14 – GA 2015
Similarly forced remain current on maintenance
Databases were functionally stable.
No Changes for several years.
The Question is: Is the new release providing
any value to you and your organization?
13. 13
Ways to Migrate from IMS to DB2
Recode the application.
Replace all IMS calls with DB2 calls.
Risk of inconsistent results.
Replicate IMS to DB2.
Great for Readers, but what application is read-
only. I want to update!
Replication delays – “Near Real Time”
Use a Transparency Layer – AKA: DL/2
Intercepts all IMS Calls
Conditionally passes them to IMS
Smoke & Mirrors
14. 14
Environmental Overview
IMS V11 running DBCTL
32 Full Function Database, 28 Primary Indexes,
85 Secondary Indexes
About 2,500 application PSB’s
About1600 Application JCL & PROC members
to change.
DB2 V10 - No Upgrade to DB2 V11 during DL/2
Migration ( Simplify the project )
CICS 4.1 & CICS 4.2
15. 15
Additional Constraints Present
in Our Environment
IMS Data Sharing.
Not a constraint, simply a complexity.
We had Replication from IMS to DB2 running
since 2001
Large number of applications as readers of the
DB2 targets of replication.
Apps had to “facilitate” a delay
There was overhead in replication: IMS User
Exit, Agents, Repositories, Additional Logging
Tasks, Post Log Merge Tasks, Apply Tasks
16. 16
Convert Replication Targets into
Views on DL/2’s DB2 Tables
Replication was one segment to one table.
At some point the light bulb came on and we
realized the targets could become views.
We were not replicating every segment or even
every field ( column ).
Some “occurs” in segments that were replicated
1 segment occurrence to 15 DB2 rows.
Occurs rows are done through a view with 14
“UNION ALL” references to same table.
Must use the key, but predicate pushes into
query and it performs great.
17. 17
Risk / Reward Tradeoff
Risks
Product won’t work ( Proof of Concept )
Implementation will take forever. ( 1 Year )
The Organizational “Naysayers”
Craziest Move: Our biggest Naysayer was assigned
as the Application Lead.
Rewards
Remove IMS from our inventory ( MLC )
Remove additional IMS Tooling ( $$$ )
Remove replication – DBA Stress Level Reduction
Simplification ( Three Major Components to One )
Position our data in a familiar relational model.
18. 18
Syncsort Defined Phases of a
DL/2 Migration Project
PHASE 1: IMS analysis and migration unit definition
PHASE 2: Define migration and test plans
PHASE 3: Map IMS databases and secondary
indexes
PHASE 4: Migrate IMS data to DB2
PHASE 5: Application testing
PHASE 6: Production cutover
19. 19
DL/2 – Project Preface & Phase 1
“Short Period of Contemplation”
December 2004 until March 2013
April 2013: IMS analysis – Done on Syncsort site
FTP them DBDLIB/PSBLIB
FTP them COBOL Copybooks for Segments
FTP code for all exit routines ( Sparse Indexes )
April 2013: Defined POC databases
21. 21
Definition – Migration Options
Option 1 – Big chunks of CHAR with key columns.
Option 2 – Most fields made CHAR columns.
Option 3 – Most fields made appropriate data type.
Data cleansing is likely.
Option 4 – Similar to Option 3, but includes mapping
segment types to different tables.
NOTE: It is possible to have any combination of these
options for your various databases to be converted.
Decision on which option to use determines the “ease
of use” later on.
22. 22
DL/2 – Project Phase 2
July 2013: Definition of Migration plan
Opted for 2 groups = 2 “steps” in each environment
Second group would also include IMS Elimination.
Two passes is similar project plan to DB2 migration
plan with CM and NFM steps.
About 50% of data ( Not an even split on databases )
Couldn’t consider all at once, and didn’t want small
groups or individual database steps to drag on
forever.
Test plans were “business as usual” in Dev/Sys/QA
Thorough test as part of DR Test.
23. 23
DL/2 – Project Phase 3
July 2013 – Syncsort prepares mapping for POC
databases and testing migration.
Lots of “Data Quality” questions
Lots of data analysis discoveries
Effects on “Migration Option Decisions”
August 2013 – On-Site POC
Several DBA’s now involved.
Minimal work by CICS system programmer.
Application specialist with extensive
IMS knowledge
A week of long days as we learned DL/2.
POC Complete – Green light for the project !
24. 24
DL/2 – Start Your Engines
Divide and Conquer
Re-link all programs to get DL/2 Stub included.
Change SCLM Build to link DL/2 Stub.
JCL Changes for BMP JCL and Procs.
Edit macro for JCL changes: about 1,600 members.
Progress Tracked Weekly.
CICS Programs Installed across the board.
DB2 Plans Changed.
PSB’s become “driver” with Static SQL in a Package.
Plans needed collid of DL/2 packages in the packlist.
Create new Bufferpools ( GBP ) for DL/2 Tablespaces
and Indexes. ( DFSVSAMP )
25. 25
DL/2 – Project Phase 4, 5 & 6
Syncsort delivers mapping.
Generate drivers.
Any issues in testing may be mapping related.
Test migration ( IMS Unload / DB2 Load ) process.
Convert Development, System Test & Quality.
Head to Production.
Quality mimicked production when production cutover
occurred. Did this to facilitate problem recreation.
Syncsort on site during both Production Cutovers.
Excellent support from Syncsort project players.
26. 26
Group Cutover Timeline
Group 1 Migration
January 29, 2014 – Test / Development
March 9, 2014 – System Test
April 30, 2014 – Quality Assurance
September 7, 2014 – Production
Group 2 & IMS Elimination Mode Migration
July 2, 2014 – Test / Development
July 16, 2014 – System Test
September 17, 2014 – Quality Assurance
October 5, 2014 – Production
27. 27
Special Processing and Issues
Data Warehouse – HD Unload processing became
DSNTIAUL Unloads.
Issue with IMS code having “SQL COMMIT” rather
that CHKP.
Ignored by IMS
Issue in IMS Elimination mode as it closes DL/2 Cursors
Unconverted developer JCL – U3303 abends.
Alignment Errors on PCB’s
DB Call against I/O PCB
Code had been wrong for years.
28. 28
Issues continued…
2 or 3 Programs with IMS call pattern not conducive to
performance.
Didn’t perform great in IMS
Performed worse under DL/2
Added relational call to replace IMS calls.
New diagnostic processes for developers.
Abend-AID, Dumpmaster
Xpeditor setup.
Maintained all doc for project on Wiki, including how-to’s
for developers.
29. 29
Reaping the rewards…
Removed IMS from our inventory ( MLC )
Removed additional IMS Tooling ( $$$ )
Removed replication – DBA Stress Level Reduction
Removed the workload of maintaining those products.
Simplification ( Three Major Components to One )
Positioned our data in a familiar relational model.
30. 30
Thanks for Attending!
That concludes my portion of the presentation.
Syncsort has some additional information to share.
Following that, there will be a time for questions.
I will gladly answer any questions related to our
implementation project.
Technical questions about the product can be
answered by Syncsort.
Thanks again for attending.
31. Typical Migration Project
Syncsort Confidential and Proprietary - do not copy or distribute
31
• Databases grouped by application/size/complexity for optimal testing
• Primarily 3 main project phases:
Database Group 1
Database Group 2
Database Group 3
TIME
Database Group n
Eliminate IMS
Database mapping usually
performed by Syncsort
Map/migrate Application testing Production cutover
32. Syncsort Migration Services
Syncsort Confidential and Proprietary - do not copy or distribute
32
• Free PSB and DBD analysis
• Proof of concept
• Mapping and migration services
• Onsite support for production cutover
33. Why the Transparency Approach?
Syncsort Confidential and Proprietary - do not copy or distribute
33
• NO application program changes are required
• LOWEST RISK migration strategy – bite-sized chunks
• REDUCED COSTS through elimination of IMS DB and BMC tools licenses
• SIMPLIFICATION of DBA support through DBMS convergence
• CHOICE of migration methods to meet differing requirements
• Simple, rapid method to reduce costs very quickly
• More measured method to meet DB2 design objectives
• A combination of both
• Simple method first, measured method later
• MINOR JCL changes (which can be made in advance)
34. DL/2 Transparent Data Migration
Syncsort Confidential and Proprietary - do not copy or distribute
34
• Eliminate IMS DB and IMS tools License costs with NO changes to
application code
• Support Legacy Modernization initiatives
• Reduce Risk as IMS DBA and Programming skills head for retirement