WAVV 2004 - VSE to Linux

Document Sample
WAVV 2004 - VSE to Linux Powered By Docstoc
					How to get from VSE to Linux

                     unlock the value of legacy
Presentation Information
• Moderator: Chris Sims
• Author: Michael Waters
• Company: Micro Focus
• Contributors: Mathias Mezger and Tom Ryan
• Presentation Title: How to get from VSE to Linux
• Presentation Abstract: Linux and VSE are a wonderful
  combination, but have you ever wondered what is involved to
  convert your VSE system to Linux. Then this is the session for
  you. Case studies of Customers who have successfully
  converted their VSE systems to Linux will be presented.
Qualifying Questions
• Do you have a portfolio of legacy COBOL applications?
• Do you have a mandate to reduce operational costs AND
  increase agility?
• Is there a risk that your legacy infrastructure will constrain your
  business options?
• Can you afford to rip and replace your legacy applications?
• Are you still running VSE/ESA?
Gartner Quotes
• "As Linux competes with Windows and Unix, the long slow
  decline of the mainframe platform becomes hard to ignore."
• "Efforts to maintain and develop mainframe applications with
  fewer programmers, while beneficial in the short term, do not
  resolve the long-term problem of skills availability for this
• "IBM mainframe enterprise less than 500 MIPS are the most
  likely to successfully migrate…"
Legacy Value
• Transactions encapsulate
  your workflow
   – CICS, DB2, and IMS
• COBOL defines your
  business rules
• Your data reflects your
Mainframe Migration Choices
• Re-engineering or “Rip and Replace”
    – High risk, time and cost
    – 74% fail or are significantly late and over budget – Standish Group
• Packaged applications
    – Change business processes or package functionality
    – Lost functionality, retraining and complex data remodeling
    – 42% never complete or are more than 50% late – Gartner
• Mainframe-based modernization
    – Low risk and minimal retraining
    – No cost savings and limited flexibility
• “Lift and Shift”
    – Low hardware and software costs
    – Low risk and minimal retraining
    – Modern, open and extensible           Lift and Shift often the best strategy
Lift and Shift Migration

Reduce Cost – Increase Agility – Minimize Risk
  Re-deploy COBOL to reduce customers’ costs while positioning
  their legacy software assets for the future, with low risk
        • Lower total cost of ownership
             – Move COBOL to Linux
        • Lower cost of integration
             – Connect COBOL with J2EE
        • Position for change
             – Service Oriented Architectures
        • Decrease risk of failure
             – Reuse skills, processes, applications, and data

 Corporations have huge investments in applications, skills and data
Lift and Shift – Phased Approach
•   Migrate applications to Linux with minimal changes
    –   Lower cost
    –   Maintain current functionality
    –   No interuption in day-to-day business operations
•   Modernize applications to take advantage of new technologies
    –   Select changes that maintain the value of legacy assets
        •   Business Logic (COBOL)
        •   Presentation Logic (Java, HTML)
        •   Communication (Web Services)
        •   Transaction Controls (J2EE – WebSphere)
    –   Reuse business functionality to provide new, modern and cross-
        platform availability to your business operations
Lift and Shift – Migration

 VSE/ESA            Linux                 Linux

 COBOL            COBOL               COBOL
 BMS              BMS                 Java / HTML
 CICS             ES/MTO              Web Services
 DB2              UDB                 UDB
 VSAM             ISAM                XML

           VSE/ESA to Linux for zSeries
Migration Considerations
• VSE/ESA applications are not always pure COBOL, CICS,
  Batch, DB2 and VSAM
• Languages other than COBOL and C may not be supported
   – Assembler is not supported
• Data sets may have to be converted
• No JES and RACF for Linux
• Third party tools may not exist on Linux
• “Evolved“ applications can have unusal personalities
Migration Considerations II
• Assembler
   – Conversion utilities from Software Migrations Limited and others
• Codeset
   – Micro Focus utilities convert data files from EBCDIC to ASCII using
     a data file layout structure to handle binary data
• Batch JCL and schedulers
   – Recommend converting JCL to Linux scripting
• Security
   – Linux security model is like UNIX, not RACF
   – Still a significant concern for Web services
   – IBM mainframe dialects are generally supported
Migration Considerations III
• Handling of numeric data-items
   – These can differ depending on the COBOL operation
   – Example:
         01 INVENTORY-FIELD.
             03 INV-FIELD-NUMERIC PIC 9.
         01 CALC-FIELD             PIC S9 COMP-3.
       *         Instructions:
         procedure division.
             IF INVENTORY-FIELD = '1'
                MOVE 'Y'     TO INVENTORY-FIELD
                MOVE 'N'     TO INVENTORY-FIELD
             TO CALC-FIELD  S0C7 might occur on mainframe

   – Resolution: Recode or use Linux COBOL Compiler directive to
     establish run time time error 163.
Migration Considerations IV
• Recent Linux conversions
   – Natural
   – Adabas
• No Linux equivalent for…
   –   PL/I
   –   CSP
   –   CA-Datacom/DB
   –   CA-IDEAL
   –   CA-IDMS (conversion utilities by Xitec Software and others)
   –   IMS
• Systems Integrators, including IBM Global Services, have
  experience migrating VSE/ESA to Linux
Case Studies
• German Bank
• Home Furnishings Retailer
• Clothing Retailer
• Direct Mail Application Provider
• Financial Application Provider
German Bank
• Original Environment
   –   1,000 programs - 50% online, 50% batch
   –   Few system routines in ASM
   –   VSAM and sequential data
   –   Several third party products
• Target Environment
   –   z/Linux, Micro Focus COBOL, Micro Focus Server (CICS)
   –   ASM programs obsolete or rewritten
   –   VSAM migrated to UDB/DB2
   –   Cancel as many third party products as possible
• Motivation
   – Reduced total cost of ownership by 50%
Home Furnishings Retailer
• Original Environment
   – z/OS, COBOL, CICS                       – 2,000 programs
   – 350 4GL programs                        – 4,000 JCL members
   – DB2, VSAM and sequential data           – Third party products
• Target Environment
   –   z/Linux, Micro Focus COBOL, Micro Focus Server (CICS)
   –   4GL programs rewritten
   –   UDB, ISAM (IDX-8) and sequential
   –   Job Control ==> BASH
   –   Cancel as many third party products as possible
• Motivation
   –   Spin-off required smaller data processing center
   –   Reduce total cost of ownership
   –   Free up budget for new application development
   –   Reduce dependency of mainframe veterans
Clothing Retailer
• Original Environment
   –   500 programs, 50% online, 50% batch
   –   1,200+ JCL members
   –   CA-Datacom, VSAM and sequential data
   –   Several third party products
• Target Environment
   – UNIX or z/Linux                    – Migrate PL/I to ??
   – Migrate CA-Datacom to UDB          – COBOL and Java mix
   – Reduce third party products        – Job Control ==> ??
• Motivation
   – Reduce total cost of ownership
   – Strategic move to e-commerce
Direct Mail Application Provider – Group 1
• Original Environment
   – z/OS and UNIX
   – Micro Focus COBOL
• Target Environment
   – z/Linux, Micro Focus COBOL
   – Source code unchanged
• Motivation
   – Satisfy customer demand (Citibank)
Financial Application – ABK Systeme GmbH
Transaction                      Development               Production                Linux Test
Performance                      (UNIX and Wintel)         (Sun, HP/UX, AIX)         (IBM zSeries)

Maximum hourly volume            0.4 million               1.6 million               19++ million
(burst speed)
Total daily volume               1.5 million               8.0 million               155++ million
Burst speed to total daily       1 to 3.75                 1 to 5.0                  1 to 8.1++
volume ratio
Source: Giga Research, a wholly owned subsidiary of Forrester Research, Inc.
Note: The test data represents real, depersonalized orders from an existing bank, run on an IBM
zSeries 900, three CPUs, three parallel I/O channels, SuSE Linux Enterprise Server

               –   A few days to port the code from Solaris to Linux
               –   Ten months from alpha to production
               –   Micro Focus COBOL used on all platforms
               –   zSeries test by IBM Lab in Böblingen, Germany
Legacy is an asset
Questions and Answers
   – Ed Anderson, Corporate Linux
        • +1 512-838-9426
        • eda@us.ibm.com

• Micro Focus
   – Michael Waters, Alliance Manager
        • +1 301-838-5259
        • michael.waters@microfocus.com