mainframe re-hosting with oracle tuxedo: accelerating cost

An Oracle White Paper Updated March 2009

Mainframe Rehosting with Oracle Tuxedo: Accelerating Cost Reduction and Application Modernization

Oracle White Paper—Mainframe Rehosting with Oracle Tuxedo: Accelerating Cost Reduction and Application Modernization

Executive Overview............................................................................. 1

Application Portfolio Analysis .............................................................. 7

Mainframe Modernization Approaches ............................................... 9

SOA Integration............................................................................. 11

Rearchitecting to a New Environment........................................... 12

Replacement with Packaged Applications .................................... 13

Rehosting or Replatforming .......................................................... 13

Automated Migration ..................................................................... 15

Oracle’s Solution for Mainframe Rehosting....................................... 16

Critical Success Factors in Mainframe Rehosting............................. 18

Preserving Business Logic and Data ............................................ 19

Meeting Performance and Other QoS Requirements ................... 24

Phased Migration and Mainframe Integration ............................... 26

Project Implementation ................................................................. 27

Further Modernization and SOA Enabling Rehosted Assets ............ 28

Further Rearchitecture of Rehosted Applications ......................... 32

Conclusion and Customer Examples ................................................ 32


Executive Overview The backlog is the number-one hurdle to effectiveness for CIOs followed by inadequate budgets and a shortage of time for strategic thinking and planning, reports “The State of the CIO 2006” survey by CIO Magazine. For IT executives managing mainframe applications, these issues are all too familiar.

After all, the average Fortune 100 company maintains 35 million lines of code in legacy applications. Literally generations of development teams have evolved these applications to support a whole array of business-critical processes. This legacy is both one of IT’s biggest assets and one of its greatest liabilities. On one hand, the two-thirds of business-critical applications that are running on a mainframe contain the invaluable business logic that handles core transactions and manages customer data. On the other hand, these organizations are locked into aging technologies. As a result, they face high operating costs, growing skills shortages, and complex integration issues, all of which serve to drive the growing backlog. Two of the biggest challenges facing today’s CIOs and CFOs are

• Unlocking the value of key business logic for integration and reuse

• Reducing mainframe operational and maintenance costs to shift more of the IT budget to innovation

Figure 1. Most companies are struggling to shift budget allocations from maintenance to innovation.

1


Tower Group and Gartner report growing interest in legacy modernization and rehosting in financial services, government, and

other IT-intensive industries.

Companies need to “…do a better job of flipping the ratio of IT spending from maintenance and infrastructure to development by migrating away from systems and applications that are increasingly costly to maintain,” recommends Tower Group1. Many Oracle customers that seek to reallocate IT spending from maintenance to innovation have modernized their legacy applications and reduced mainframe infrastructure costs by rehosting mainframe applications to modern application platforms running on open systems. The Oracle mainframe rehosting solution helps enterprises and government organizations to substantially reduce costs, constrain MIPS growth, and accelerate SOA enablement—and accomplish all this without affecting business operations, increasing risk, or sacrificing quality of service (QoS) for their customers and business users.

Tower Group and Gartner report growing interest in legacy modernization and rehosting in financial services, government, and other IT-intensive industries. Why is legacy modernization garnering such great interest now? The most visible factors include:

• Organizations that depend on IT for competitive differentiation, customer responsiveness, and higher margins cannot afford2 the innovation required to support these goals when two-thirds of the IT budget goes toward maintenance, leaving only 30 percent to invest in new development. Growing mainframe costs, including IBM and 3rd party software, make MIPS containment and TCO control a higher priority now than ever before.

1 2007 Insurance Industry Perspectives: Business Drivers, Strategic Responses, Technology Initiatives, Tower Group, 11/2006. 2 Tower Group emphasizes that “technological obsolescence continues to impact the industry as firms deal with aging technology platforms that house core data and processes. CIOs are forced to weigh the cost of maintenance on these legacy systems against the cost of building or buying new systems as maintenance becomes a larger financial burden.” The 2007 Top 10 Business Drivers, Strategic Responses, and IT Priorities for Asset Management Firms, Tower Group, 10/2006.

2


• Risks related to vendor lock-in and graying skills3 have risen, while the risks of migrating from the mainframe have been significantly reduced by the mainframe-class capabilities of the key platform and middleware technologies that have been leveraged to support a number of visible, successful rehosting and modernization projects in recent years.

• IT consolidation is leading many enterprises to focus on platform cost optimization as part of their infrastructure strategy, which then becomes the driver for a number of rehosting projects. Older pre-System z IBM mainframes have reached an end of life, forcing customers to migrate or face expensive replacements. Non-IBM mainframes and mainframe software, such as Unisys ClearPath, CA’s Datacom and IDMS, Software AG’s Natural and Adabas, and Bull GCOS present significant risk of obsolescence.

• Modern IT best practices, such as enterprise use of Service-Oriented Architecture (SOA) and Application Portfolio Management (APM) provide a migration path for legacy application assets and offer the best way to maximize their value. According to Tower group, “Ultimately, the analysis will result in one of two recommendations:

• Renovation if Web services tools can make applications work better together at a reasonable cost, or

• Replacement if the operational pain (from the perspectives of both functionality and IT maintenance) requires a clean break with old applications.”

For many organizations, mainframe modernization has become a matter of “how,” and not “if.” Oracle, together with its services and hardware platform partners, provides a proven mainframe modernization approach that many have leveraged to successfully

3 Accelerating retirement of business-knowledgeable and mainframe-skilled IT staff is increasing the risk to sustainability of mainframe applications over the next 5-10 years. The Business and Technical Drivers of Mainframe Modernization, Gartner, 12/2006

3


The Oracle mainframe rehosting solution helps enterprises and government organizations to substantially reduce costs,

constrain MIPS growth, and accelerate legacy modernization,

migrate selected applications and complete mainframe environments, ranging from a few hundred to over 10,000 MIPS. Combining the power of Oracle Database and Fusion Middleware with Oracle Applications enables a broad range of modernization solutions

Oracle Database grid is the cornerstone of the Data Management layer. For custom applications, the application infrastructure layer consists of Oracle Tuxedo and WebLogic which host the business logic—Tuxedo hosts rehosted COBOL and C/C++ and WebLogic hosts Java/J2EE for extended services, new components, etc. Both provide high degree of SOA capabilities and integrate with the Oracle SOA Suite technologies that form services infrastructure layer, and standards-based service networks. These technologies together enable customers to rehost, SOA enable, rearchitect, extend, and reuse the critical applications residing on the mainframe.

Figure 2. Oracle technology for modernization and SOA enablement of mainframe applications

4


Oracle solution delivers an optimal target environment powered by rock-solid Oracle Database and Oracle Tuxedo as transaction processing (TP) infrastructure and COBOL, C, and C++ application server that closely matches mainframe requirements:

• Support for hosting COBOL or C/C++ business logic and IBM CICS and IMS-equivalent capabilities

• Mainframe-class reliability, availability, performance, and scalability (RASP) for which Oracle Database and Tuxedo are well known in the industry

• Extensibility with Web Services, J2EE integration, enterprise service bus (ESB), business process management (BPM), portal, and enterprise performance management (EPM) technologies to enable easy integration of rehosted applications into modern SOAs without rewriting to Java or .Net.

• End-to-end transaction and service monitoring to support 24/7 operations management with reliability and manageability characteristics similar to mainframe installations

• Robust clustering technologies for high availability and fail-over capabilities within a data center or across the world

Leveraging these technologies, Oracle System Integration partners deliver predictable, cost-effective projects using risk-mitigation methodologies and automated tools honed over numerous migrations to address a complete range of online, batch, and data architectures. The industrialized methodologies support all phases of a migration project from discovery and analysis to conversion, integration, testing, and switch-over.

Taken together, all of these capabilities have led to a large and growing number of mainframe migration projects which successfully rehosted large, multi-generational mainframe applications, while avoiding expensive and unpredictable rewrites. The long-term investment in critical business logic and data was preserved without risking

5


Oracle, together with its services partners, provides proven, comprehensive solutions that many customers have leveraged for

a successful migration projects that preserve the long-term investment in critical business logic without risking business

operations or sacrificing mainframe QoS characteristics.

business operations or sacrificing Quality of Service. Customers using Oracle solution to rehost mainframe applications have experienced the following benefits.

• Reduced or eliminated mainframe maintenance costs and defer upgrade costs, saving CIOs 50-80 percent of their maintenance budget

• Increased productivity and flexibility in IT development and operations, protecting long-term investment through application modernization

• Speeded up and simplified application integration by providing an SOA framework for their business-critical applications that used to run on the mainframe—without sacrificing the transactional integrity and high performance expected by the users.

This white paper explores the key success factors learned from many of these migration projects. It also presents proven transformation architecture for rehosting mainframe applications and data, describes SOA integration options and considerations when SOA-enabling rehosted applications, and explains project methodology commonly used for risk mitigation in these projects.

6


7

Application Portfolio Analysis

The first step in approaching mainframe migration and application modernization is to identify and analyze your application assets. The Aberdeen Group estimates that there are “… $2 trillion worth of mainframe applications in corporations that house approximately 70 percent of all critical business logic and data.” Some of these applications have been part of the IT landscape for years; others are fairly recent and are still changing rapidly. It is more difficult than ever for IT teams to keep track of all mainframe application assets in their portfolio, their dependencies, and ongoing changes, in part because tactical problems often overwhelm their resources and delay strategic planning.

IT departments can use application portfolio management (APM) methodologies to review their application portfolio, determine the business value of the various applications, find ways to reduce complexity and optimize utilization of their key resources through consolidation of platforms, tools, and languages, and find ways to shift resources from maintenance to innovation in order to better align IT with the business. APM helps to provide a strategic framework for these goals, including guideposts for modernization. In “Is It Time for a Mainframe Makeover?,”4 author Mike Gilbert, Ph.D. recommends an approach to APM that scores applications on two key dimensions—strategic importance and operational efficiency—using the key factors and metrics in the table below.

TWO DIMENSIONS OF APPLICATION SCORING IN APM

STRATEGIC IMPORTANCE OPERATIONAL EFFICIENCY

• Contribution to revenue

• Contribution to profit margins

• Importance in differentiating the business

• Importance for compliance

• Cost per transaction

• Quality of service (QoS)

• Flexibility to adapt to new business needs

• Use of contemporary technologies

Modernization efforts focused on the applications with high strategic importance and low operational efficiency produce the highest return, as the value of the application can be further enhanced while raising its operational efficiency. For these applications, reducing the cost per transaction, increasing their business agility, and integrating contemporary technologies can offer the biggest payback. It is important to narrow down the applications to be modernized because, 4 “Is It Time for a Mainframe Makeover?” z/Journal. http://www.zjournal.com/index.cfm?section=article&aid=774 (accessed October 1 2007).


as Dr. Gilbert says, “modernization occurs over a series of interrelated projects, possibly lasting several years.” For organizations with strategic reliance on IT and high IT complexity—such as banks, insurers, asset management firms and other financial services, as well as public sector organizations with significant financial and transactional workloads, such as the Social Security, Tax and Revenue, Unemployment, and Health Insurance, Dr. Gilbert recommends an “steady-state” modernization approach that becomes an integral part of how IT is governed—modernization as a constant focus rather than a one-off activity. Once the appropriate applications have been identified, the analysis can extend beyond the value provided by the applications today and their related maintenance and operating costs to address the following questions:

GUIDE TO APPLICATION-SPECIFIC ANALYSIS

INCREASING APPLICATION VALUE

THROUGH MODERNIZATION

REDUCING COSTS AND INCREADING

BUDGET FOR INNOVATION

MINIMIZING THE RISK AND SCOPE OF

CHANGE

• What’s the application’s current

value based on strategic importance

criteria?

• Does it support your current

business needs or requires a major

revamp?

• Will modernization make it easier to

reuse the application’s capabilities

and to integrate it with other

systems?

• If you unlock the application from its

host environment, can its services

and business rules be leveraged for

new, strategically important uses?

• How often do you need to implement

process changes and regulatory

mandates?

• If you do nothing, what’s a potential

impact on competitiveness,

compliance, or other critical areas in

a few years?

• What is the application’s current

TCO—including the annual cost of

current infrastructure, variable

workload-based and other software

charges, and operational resources

used?

• Can shifting to a new platform &

software stack significantly reduce

TCO?

• Would a platform shift reduce

dependence on scarce mainframe

skills in favor of using more widely

available skills?

• How viable is the new platform and

software stack in your organization?

How much of it already exists and is in

use?

• How will business and technology

trends affect your platform strategy in

the 3-5 years?

• What are the external deadlines or

internal pressures driving your

timeframe for migration?

• How can you analyze and

document key business rules and

dependencies?

• How can the business logic be

preserved with the minimally

invasive changes?

• How to encapsulate and extend

core COBOL logic with Web

services and/or an ESB?

• How can you transform UI

elements to Web-based Java

technologies?

• How can you simplify integration

with newer applications for greater

reuse of key processes?

• How can you ensure that the new

and old infrastructures work

together during the phased

transition?

• What are the critical change

requests you need to address as

part of the modernization process?

A macro-level analysis based on these questions provides a good business context for modernization discussions. It can be combined with the micro-analysis that examines applications at the program and data schema level to track dependencies, generate complexity and maintainability metrics, find areas of dead code, and discover key interfaces. The

8


combination of macro- and micro-analysis will provide the context and specific application data that helps to build a modernization roadmap by combining rehosting to a new platform, restructuring selected components for SOA-readiness, and/or rearchitecting the overly complex and difficult to maintain modules for increased agility.

In addition, the analysis process has to address the issues of increasingly scarce mainframe skills. Scarcity of such skills has become an increasingly bigger factor in the decision about mainframe modernization because generational changes make the skill shift inevitable, and delaying only increases the risk to your application portfolio. Gartner has predicted that “by 2008, organizations that have not developed a mainframe modernization strategy risk losing the opportunity to leverage this portfolio due to a decline in skills and knowledge of these systems (0.7 probability).”5

The information you gather with the application portfolio analysis outlined above allows your company to establish a strategic IT modernization roadmap and to evaluate alternative approaches to modernization from a business and IT perspective. The next chapter discusses these alternatives in more detail highlighting their risks and benefits.

Mainframe Modernization Approaches

The challenges of legacy mainframe applications and systems—pressure to contain or reduce mainframe costs, accelerating retirement of mainframe-skilled staff, and detrimental impact to the backlog and business agility—are well known in most IT organizations, and CEOs and company boards are increasingly recognizing their impact. What strategies are companies using to address them? What are the primary approaches available and what are their tradeoffs? We highlight below the options that organizations consider and examine their challenges and benefits.

Some of these approaches leave the applications on the mainframe, while looking to increase their agility or reduce costs. Other approaches migrate the applications off the mainframe to a UNIX, Linux, or Windows environment, leveraging options that range from least disruptive to a complete rewrite or replacement. The information gained through application portfolio analysis and the organization’s objectives can be used to evaluate multiple approaches and options. These objectives often include substantial cost reduction, increase in business agility and IT flexibility, time constraints (internal or external), and other factors.

5 “Leverage Your Mainframe Applications With SOA.” Gartner. http://www.gartner.com/DisplayDocument?ref=g_search&id=486559&subref=simplesearch (accessed October 25, 2005).

9


Before we delve into modernization options, we should also address the possibility of doing nothing or outsourcing the mainframe to a third party. While not specifically addressing modernization needs, they do come into the conversation as potential alternatives.

Doing nothing does not guarantee an unchanging budget since costs can grow rapidly when an upgrade or mainframe technology refresh is required. The long-term effect of doing nothing could also be costly in other ways. It leaves the many legacy issues discussed earlier to the mercy of ad-hoc solutions, often at substantial extra cost and risk to the business. There is a very real concern that companies will not be able to sustain support of these business-critical applications as the population of workers with mainframe skills ages and is not being resupplied.

The promise of outsourcing mainframe operations or application maintenance or both is that it lowers costs initially. However, the ongoing management cost and the cost of any required changes over time are often an unknown factor. Any changes required in the configuration, the applications, or other aspects of the mainframe environment could add significant incremental costs to the outsourcing bill. Issues such as compliance auditing in an outsourced situation and the risk created by sharing critical infrastructure, which is how the initial cost reductions are achieved, may create yet additional, often unexpected costs.

The risks associated with outsourcing relate to having a third party assume the management and control of your business-critical applications and their supporting infrastructure. How stable is the shared infrastructure your outsourcing vendor provides? Will the vendor continue to provide the responsive service that your company needs? What if your company’s needs change—can the vendor adapt?

While the long-term cost reductions of outsourcing are uncertain, one thing is certain—you are still locked into the legacy environment, with no path to attain the business benefits of application modernization. There is the risk from your competitors with more agile systems growing market share at your expense. A system that lacks the agility to meet regulatory compliance requirements can also create greater risks. When companies avoid or delay legacy modernization, they are at high risk to lose their competitive edge in the global marketplace.

The diagram below illustrates the range of five true modernization options customers consider to meet their cost reduction and business agility objectives. Of these five options, SOA Integration is the one that keeps the applications in the mainframe environment, but uses adapters and Web Services to open them up for integration into the enterprise’s SOA. The other four options are used to move the mainframe application to an open systems environment—UNIX, Linux (including z/Linux running on mainframe hardware), and Windows. Rehosting and automatic migration tend to be the least invasive and protect the business logic and data of the application, while delivering higher ROI and faster results. Replacing applications with COTS solutions or rearchitecting to a new custom application tends to maximize the use and the benefit of modern technologies (BPM, Rules, etc.), but involves the most changes and the highest risk.

10


Figure 3. Approaches to mainframe modernization

Real world projects often straddle these boundaries and mix multiple approaches to achieve the most effective results. For example, a company may start by replacing standard processes such as HR or payroll with packaged applications, while rehosting remaining custom processes intact, leveraging compatible platforms and automatic migration to achieve rapid cost savings. These cost savings can then be used in part to fund additional rearchitecture phases to integrate the rehosted components using SOA integration capabilities of the target environment, while selectively transforming some of the rehosted modules requiring frequent updates into BPM-driven processes. Oracle modernization experts can help you analyze the broad range of approaches and chart the modernization course to optimally meet your application requirements and business objectives.

SOA Integration

One approach to modernizing mainframe applications is to extend them in place by changing their user interface (UI), adding support for Web services protocols, and exposing new, higher level APIs. This may require upgrading some of your system software or installing 3rd party integration products on the mainframe. For example, to SOA-enable IBM’s CICS Transaction Server applications, you must upgrade to CICS Transaction Server V3.1 or 3.2 and may also have to install components of the IBM WebSphere application server on the mainframe. Also, IT will have to adapt development tools, expand test environment and procedures, and upgrade operations and management processes and tools. It is important to benchmark the SOA integration to determine if the additional processing load on the system will require additional resources—such as increased memory or CPU capacity—to provide adequate response time and throughput. One upgrade option is to leverage IBM’s special-purpose mainframe processors, such as IBM System z Application Assist Processors (zAAPs) for Java and IBM System z Integrated Information Processors (zIIPs) for DB2, which are priced lower than the general purpose processors. However, the traditional applications running in CICS or IMS can’t take advantage of these processors, which limits the potential cost savings.

11


SOA integration approach wraps key application interfaces in services, and integrates it into the SOA. This largely leaves the existing application logic intact, minimizing changes and adding risk only to those components that needed restructuring work to become SOA-ready. While the interfaces are modernized without subjecting the core application components to a lot of change, the high costs and various legacy risks associated with the mainframe platform remain. In addition, the performance and scalability of the new interfaces needs to be well-specified and tested, and the additional load they place on the system should be included in any planned capacity upgrades, potentially increasing the overall costs.

Rearchitecting to a New Environment

Rearchitecting is based on the proposition that legacy applications contain invaluable business logic and data relevant to the business, and that these assets should be leveraged in the new system, rather than throwing it all out to rebuild from scratch. This contrasts with rewriting, where the existing assets are used only to formulate the requirements, which are then used as a basis for writing the new application for the new target environment. Rearchitecting often includes rewriting or refactoring of the legacy code to modern languages like Java. But since the modern IT paradigm elevates some of this logic above the code using declarative models supported by BPM tools, ESBs, business rules engines, data integration and access solutions, some of the original technical code can be replaced by these middleware tools to achieve greater agility. Thus, rearchitecting focuses on recovering and reassembling the application logic relevant to business from a legacy application, while reducing the technology-specific code.

The rearchitecting approach can migrate the application off the mainframe and restructure it using all of the modern software tools and capabilities at our disposal. However, this very process of restructuring the application, and essentially rebuilding it using knowledge and business rules mined from existing code, introduces its share of risks. How can we ensure that the new application maintains functional equivalence and operational characteristics of the original? Can we meet the performance and scalability requirements not only of the current environment, but future growth needs as well? Can we deliver the new application within the time and budget constraints agreed to at the beginning of the project? The older the application, the larger its scope and volume of code, and the fewer original developers available—the higher these risks may be.

Undertaking a change from one language or development paradigm to another shouldn't be undertaken lightly—the investment and risks need to be well understood and justified. It can be the right approach for more recent applications that have narrow focus and good current documentation, particularly if the original development team is able to guide the process. In larger and older applications selective rearchitecture may be a better approach, focusing on components that require frequent maintenance changes but are difficult to maintain, because of poor structure and layered changes. The payback on this rearchitecture investment will come from reducing the cost of future maintenance. Similarly, components that need significant

12


functional changes to meet new business requirements can benefit from substantial productivity increase after rearchitecture to a more modern development framework with richer tools to support future changes. The payback comes from greater business agility and time-to-market improvements.

On the other hand, well-structured COBOL components that do not require extensive changes to meet business needs will show very little return from rearchitecture investment. Leaving them in COBOL but rehosting to a modern, extensible platform saves significant costs that can be invested elsewhere, reduces risk, and shortens payback time. These considerations can help to optimize ROI for medium-to-large modernization projects encompassing hundreds or thousands of components with millions or tens of millions lines of code.

Replacement with Packaged Applications

The option of replacing applications with packaged solutions exemplifies the “buy” approach to application strategy rather than “build”. By relying on a packaged solution from an application vendor the company can focus IT resources on other critical activities and let the software vendor support and enhance the solution. How well this approach works depends on the availability of packaged solutions for your specific industry and for your particular application needs. Applications like HR management or CRM may well be a good way to shift internal efforts to more strategic aspects of the business, relying on the packaged solution vendor for support of core business processes.

However, in some industries companies rely on unique business processes to gain competitive advantage. Some companies have as much as 80 percent of the applications they use custom written, using the technology to differentiate themselves in the market. If a packaged application does not support all of your requirements, consider supplementing it by rehosting key custom components and integrating them with the packaged application. Oracle Fusion Middleware is designed to seamlessly integrate Oracle Applications with custom-built components.

An additional consideration in replacing custom applications with packaged that needs to be factored into the overall transition costs is a need to customize the application’s configuration to your specific needs and train end users on the new application.

Rehosting or Replatforming

Rehosting enables companies to preserve mainframe applications, business logic, and data intact and to migrate them to a lower-cost platform by using a mainframe-compatible software stack that natively supports COBOL and provides the execution environment compatible with mainframe’s IBM CICS or IMS TM for online applications or JES2/JES3 batch environments. Rehosting protects legacy investment by relying on a mainframe-compatible software stack to minimize any changes in the core application, and preserve the application's business logic intact,

13


while running it on an open-system platform using more flexible and less expensive system infrastructure and more extensible application architecture using:

• Relational DB grid and open transaction processing platform to scale on demand and optimize the infrastructure costs based on need.

• Web services and ESB technologies for standards based integration

• BPM for orchestrating and running business processes that combine application services and human interaction

• Service metadata repository for lifecycle governance of the services in their SOA

• J2EE or .Net to extend the application with new components that easily integrate with the existing, proven COBOL services developed over the years

The right target software stack will support these requirements and also maintain the performance and reliability of the mainframe. Oracle Tuxedo is the leading rehosting application platform that is known for its mainframe-class QoS characteristics. Its broad extensibility keeps open customer’s options for SOA enablement and rearchitecture by providing support for Web Services and ESB interfaces for rehosted components. Using an extensible platform open to integration with J2EE components, BPM-based processes, and other key SOA tools means you can start rearchitect selected components at will, while maintaining strong integration with the rehosted services running the bulk of the business logic.

This approach allows you to significantly lower the system’s TCO and carries substantially lower risk than attempting to migrate a large and complex application to a different language or architecture. And in combination with Automated Migration for Assembler, 4GLs (e.g., Natural, CA Gen, ADSO, IDEAL, etc.) or languages like PL/I, you can eliminate the need for a rewrite.

Rehosting is the optimal choice when you need to preserve the core business logic or data of the mainframe application because the company continues to rely on the business processes they support. If the value of the business processes or the application’s support for them falls significantly short of your needs, the application’s value diminishes, and it’s probably best to look at other options. But if the application’s value is significant, then consider rehosting the application to lower or eliminate the mainframe cost. Rehosting eliminates the need to maintain antiquated skills for mainframe maintenance, and will modernize the application platform and interfaces—making it easier to extend functions and to integrate with other systems.

Companies have rehosted applications successfully from mainframes with processor speeds ranging from less than 100 MIPS to over 10,000 MIPS. Rehosted mainframe environments include IBM zSeries mainframes, Unisys ClearPath mainframes, Bull GCOS mainframes, and others.

14


Automated Migration

Often used to complement rehosting or rearchitecting, automated migration addresses the critical need for environments or languages that are not available or not easily supported on open systems. An obvious example is IBM Assembler, which can not be executed without the IBM mainframe hardware architecture, but can be converted to portable COBOL. Another example is PL/I, which is not well supported on open systems, but can be converted into C/C++. Such migrations are considered automated only if at least 80 percent of the transformation can be handled by tools rather than manually. In order to reach this degree of automation, the migration or transformation process must be “algorithmic” in nature and not require the injection of human intelligence into the transformation process (as is needed for rearchitecting). As a general rule, such migrations are most successful when there is a well-defined mapping between the source and target architectures.

This approach is most often used to migrate 4GL environments, such as NATURAL, IDEAL, ADSO, and PowerBuilder into rehosted COBOL or to Java. These 4GLs typically consist of both a language and a runtime environment that provides additional capabilities in support of the executing programs. These fourth-generation languages were created in an attempt to move away from the complexity of third-generation languages and environments such as COBOL, PL/1, CICS, and IMS/DC by using a runtime environment to “abstract away” from legacy environments. As a result, these 4GL environments are somewhat more modern architecturally than 3GL environments. This means that they do not suffer as much from the architectural issues that make automated migration to Java virtually impossible for 3GL environments such as COBOL/CICS.

Legacy applications written in 3GL languages such as COBOL and PL/1 make use of legacy database and file systems such as VSAM, IMS/DB, ADABAS, IDMS and Datacom. They retrieve data by issuing specific API calls embedded within the program code. Given a data model mapping from the legacy database or file formats to relational tables it is possible to use Automated Migration to automatically remove the calls and replace them with SQL—leaving the rest of the code alone. Additionally, the same data model mapping can be used to migrate the actual data from the legacy database or file system to a relational database.

Another typical use of automated migration is to convert JCL batch jobs into an equivalent framework using scripting (e.g., Korn shell or Perl), while using a batch functions library to preserve the overall structure of the JCL job stream.

The clear advantages of automated migration are speed and consistency. Since a computer carries out the process, it can be done quickly and consistently, and can even be repeated on a more recent copy of the source code in order to include ongoing changes. This process will be done the same way every time, so although automated migration is more invasive than either SOA integration or rehosting, the level of risk and required testing are much lower than with any manual effort.

15


The disadvantage of automated migration is that only algorithmic transformations can be made. If the goal of a modernization effort is to make major change to architecture and application design, automated migration will not work.

Oracle’s Solution for Mainframe Rehosting

For many organizations mainframe modernization has become a matter of ‘how’, not ‘if’. Numerous enterprises and public sector organizations choose rehosting as the first tangible step in their legacy modernization program precisely because it delivers the best ROI in a fastest possible manner, and accelerates the move to SOA enablement and selective rearchitecture. Oracle together with our services partners provides a comprehensive rehosting-based modernization solution that many customers have leveraged for a successful migration of selected applications or complete mainframe environments ranging from a few hundred MIPS to well over 10,000 MIPS.

Two key pillars support successful rehosting projects:

• Optimal target environment that lowers the total cost of ownership (TCO) by 50-80 percent and maintains mainframe-class quality of service (QoS) using open, extensible, SOA-ready, future-proofed architecture

• Predictable, efficient projects delivered by our SI partners with proven methodologies and automated tools.

Rehosting—also known as replatforming—enables companies to preserve mainframe applications, business logic, and data

intact and to migrate them to a lower-cost platform.

Optimal target environment provided by Oracle is powered by proven open systems software stack leveraging Oracle DB and Oracle Tuxedo for a rock-solid, mainframe-class transaction processing (TP) infrastructure closely matching mainframe requirements for online applications, which include:

• Mainframe-compatible Transaction Processing: Support for IBM CICS or IMS TM applications in native COBOL or C/C++ language containers with mainframe-compatible TP features

• Mainframe-class performance, reliability, and scalability (RASP) provided by Oracle Real Application Clusters (RAC) and Oracle Tuxedo’s Domains clustering for load-balancing and high availability despite failure of individual nodes or network links

• Workload and System Management: End-to-end transaction and service monitoring to support 7x24 operations management provided by Oracle’s Enterprise Manager and Tuxedo System and Application Monitor.

16


• SOA Enablement and Integration: Extensibility with Web Services (via SALT), J2EE integration (WLS/WTC), Enterprise Service Bus (OSB), Portal (WebCenter) and BPM technologies (BPEL PM) to enable easy integration of rehosted applications into modern enterprise SOA.

The diagram below shows conceptual mapping of mainframe environment to compatible open systems infrastructure used for rehosting mainframe applications.

Figure 4. Transforming mainframe architecture to open systems

The stack on the left shows a typical mainframe application environment—starting with CICS or IMS TM transaction processors to host COBOL business logic, sometimes 4GLs, a variety of prerelational and relational databases, JCL-based batch environment, and 3270/BMS-based user interfaces. When rehosting this environment to open systems we leverage Oracle Tuxedo to host legacy COBOL or C business logic and provide CICS or IMS-equivalent transaction processing services. Both relational and non-relational data is migrated to Oracle DB, sometimes supplemented with an emulation layer to minimize code change. Batch environment can be migrated using emulation or converted to a scripting framework, and supplemented with required utilities and job schedulers, including Oracle Scheduler. The user interface can remain as tn3270 terminal emulation or, as many customers prefer in order to eliminate the high cost of tn3270 emulators, Web UI automatically generated from BMS or MFS screen maps.

These rehosting projects are well understood and have been successfully done with systems as large as 12,000 MIPS, tens of thousands of lines of code and hundreds of terabytes of data. Oracle SI partners and specialized modernization firms that participate in Oracle Modernization Alliance have proven methodologies and industrialized tools to ensure these projects are done

17


with low risk and delivered on time and on budget. They use risk-mitigation methodologies and automated tools honed over numerous projects to address a complete range of Online, Batch, and Data architectures and the various technologies on which they depend. These project methodologies and automated tools that support them encompass all phases of a migration project shown in the table below.

REHOSTING METHODOLOGY: PROJECT PHASES

• Discovery and Preliminary Assessment

• Application Asset Cataloguing and Analysis

• Application and Data Conversion (pilot or entire application portfolio)

• System and Application Integration

• Test Engineering

• Regression and Performance Testing

• Education and Training

• Operations Migration and Switch-Over.

Combining a proven target architecture stack well-matched to the needs of mainframe applications with mature methodologies supported by automated tools has led to a large and growing number of successful rehosting projects. There’s a rising interest to leverage the rehosting approach to mainframe application modernization, as a way to get off a mainframe fast and with minimal risk, in a more predictable manner than a broad rearchitecture of large, business-critical applications evolved over a long term and multiple development teams. Rehosting based modernization approach preserves an organization’s long term investment in critical business logic and data without risking business operations or sacrificing the QoS, while enabling customers to:

• Reduce or eliminate mainframe maintenance costs and/or defer upgrade costs, saving Customers 50-80 percent of their annual maintenance and operations budget

• Increase productivity and flexibility in IT development and operations, protecting long-term investment through application modernization

• Speed up and simplify application integration via SOA without loosing transactional integrity and high performance expected by the users.

Critical Success Factors in Mainframe Rehosting

Companies considering rehosting some mainframe applications have to address a range of concerns, which usually can be summarized by the following questions:

• How to preserve the business logic of these applications and their data?

• How to ensure that rehosted applications continue to meet performance requirements?

18


• How to maintain scalability, reliability, transactional integrity, and other QoS attributes in an open-system environment?

• How to migrate in phases, maintaining robust integration links between migrated and mainframe applications?

• How to achieve predictable, cost-effective results and a low-risk project?

Meeting these challenges requires a versatile and powerful application infrastructure—one that natively supports key mainframe languages and services, enables automated adaptation of application code, and delivers proven, mainframe-like Quality of Service on open system platforms. For rehosting to enable broader aspects of the modernization strategy, this infrastructure must also provide native Web Services and ESB capabilities to rapidly integrate rehosted applications as first-class services in an SOA.

Equally important is a proven, risk-mitigation methodology, automated tools, and project services specifically honed to address adaptation of application code and data, supported by cross-platform test engineering and execution methodology, strong system and application integration expertise, and deep experience with operations migration and switch-over.

Preserving Business Logic and Data

Applications developed to run critical business processes have evolved over many years. They contain invaluable business logic and data models that are often at the heart of the organization’s mission. Preserving this business logic and data is a first critical requirement in mainframe rehosting. We examine this issue in detail for online applications, batch jobs, and data.

Online TP Environment

The rehosting approach depends on a mainframe-compatible transaction processing and application services platform supporting common mainframe languages, such as COBOL and C, which preserves the original business logic and data for majority of mainframe applications and avoids the risks and uncertainties of a rewrite. A complete rehosting solution provides native support for TP and Batch programs, leveraging an application server-based platform that provides container-based support for COBOL and C/C++ application services and TP APIs similar to IBM CICS, IMS TM, or other mainframe TP monitors.

Oracle Tuxedo can run most COBOL/CICS applications unchanged, provide mainframe class reliability, and scale to handle

over 100,000 transactions per second with sub-second response time.

Oracle Tuxedo is the most popular TP platform for open systems, as well as leading rehosting platform that can run most of mainframe COBOL and C applications unchanged. Tuxedo’s container-based managed process execution framework combines common application server features, including health monitoring, fail-over, services virtualization and dynamic load balancing critical to large scale OLTP applications together with standard TP features, including

19


transaction management, advanced queuing, and reliable coordination of distributed transactions (a.k.a. 2-Phase Commit, XA, and Sync Level 2). It provides the highest possible performance and scalability and has been benchmarked against a mainframe at over 100,000 transactions per second with sub-second response time.

Figure 5. Oracle Tuxedo and its surrounding ecosystem

Oracle Tuxedo supports common mainframe programming languages, i.e., COBOL and C, and provides comprehensive TP features compatible with CICS and IMS TM, which makes it a preferred application platform choice for rehosting CICS or IMS TM applications with minimal changes and risks. The diagram above shows Oracle Tuxedo and its surrounding ecosystem of SOA, J2EE, ESB, CORBA, MQ, and Mainframe integration components.

In Tuxedo environment, COBOL or C business logic remains unchanged. The only adaptation required is through the transparent mapping of CICS APIs (e.g., CICS EXEC calls) to equivalent Tuxedo API functions. This mapping is achieved by combining a preprocessor and a compiler to produce the executable components, supported by a CICS or IMS TM API emulation library implemented on Tuxedo platform. The automated nature of the conversion process and comprehensive coverage provided by the library ensures that most CICS COBOL or C programs are easily transformed into Tuxedo services.

The diagram below depicts a target rehosting architecture for a typical mainframe OLTP application. In a rehosted application CICS transactions become Tuxedo services, registered for processing by Tuxedo server processes instantiated in a single machine or across multiple machines in a domain or multiple domains (functional grouping of servers.) The services are called by a broad range of front-end clients, including tn3270 emulators, Web UI, Java, .Net, or

20


by other services in case of transaction linking. Deferred transactions are handled by Tuxedo’s queuing facilities, which provides in-memory and persistent queuing services. In the diagram below, 3270 BMS maps have been converted into JSP/HTML (using automated migration approach and tools) and are managed by servlets that connect with Tuxedo services via Oracle Jolt, WTC, or SALT.

Applications using mainframe 4GLs or languages such as PL/I or Assembler can be converted to COBOL or C/C++. The adaptation of CICS or IMS TM API calls is automated through a mapping layer, which allows the development team to continue to maintain the applications in a familiar language, TP, and data environment. For more significant extensions and new capabilities, you can incrementally leverage Tuxedo’s own APIs and facilities or leverage a tightly-linked J2EE environment provided by WebLogic Server, and even make Web Services calls—the optimal extensibility options depend on application needs, availability of Java or C/COBOL skills, and other factors.

The following table shows the most common CICS facilities and the Tuxedo API or features that provides the same functionality:

MAPPING IBM CICS FACILITIES TO ORACLE TUXEDO

FEATURE OR ACTION CICS VERB TUXEDO API OR FEATURE

Communications Area DFHCOMMAREA Typed Buffer

Transaction Request LINK tpcall

Transaction Return RETURN Tpreturn

Transfer Control XCTL tpforward

Allocate Storage GETMAIN tpalloc

Queues READQ / WRITEQ TD,TS tpenqueue / tpdequeue

Begin new transaction START TRANID /Q and TMQFORWARD

Abort transaction ISSUE ABEND tpreturn TPFAIL

Commit or Rollback SYNCPOINT / SYNCPOINT ROLLBACK tpcommit / tpabort

21


Figure 6. Oracle architecture for rehosting online TP applications with 3270/BMS green screens

Oracle Tuxedo provides a platform for deploying, executing, and managing COBOL and C rehosted transactional applications requiring any of the following OLTP and infrastructure services:

• Native, compiler-independent support for COBOL, C, or C++

• Rich infrastructure services for managing and scaling diverse workloads

• Feature-set compatibility and inter-operability with IBM CICS and IMS/TM

• 2-Phase Commit (2PC) for managing transactions across multiple domains and XA-compliant resource managers (databases, message queues, etc.).

• Guaranteed inter-application messaging and transactional queuing.

• Transactional data access (using XA-compliant resource managers) with ACID qualities.

• Services virtualization and dynamic load balancing.

• Centralized management of multiple nodes in a domain, and across domains

• Communications gateways for multiple traditional and modern communication protocols.

• SOA Enablement through native Web Services and ESB integration.

An important aspect of the mainframe environment is workload monitoring and management, which provides information for effective performance analysis and capabilities that enable mainframe systems to achieve better throughput and responsiveness. Oracle Tuxedo System and Application Monitor (TSAM) provides similar capabilities, including ability to

22


• Define monitoring policies and patterns based on application requests, services, system servers, such as gateways, bridges, etc., and XA-defined stages of a distributed transactions,

• Define SLA thresholds that can trigger variety of events within Tuxedo event services, including notifications, instantiation of additional servers, etc.

• Monitor transactions on an end-to-end basis from a client call through all services across all domains involved in a client request

• Collect service statistics for all infrastructure components, such as servers and gateways

• Detail time spent on IPC queues, waiting on network links, and time spent on subordinate services.

TSAM integrates with Oracle Enterprise Manager and provides a built-in Web-based management and monitoring console and an open framework for integration with 3rd party performance management tools.

Batch Jobs

Mainframe batch jobs are a response to a human 24-hour clock on which many businesses run. It includes beginning-of-period or end-of-period (day, week, month, quarter) processing for batched updates, reconciliation, reporting, statement generation, and similar applications. In some industries, external events tied to a fixed schedule such as intra-day, opening or closing trade in a stock exchange, drive specific processing needs. Batch applications are an equally important asset, and often need to be preserved and migrated as well. The batch environment uses Job Control Language (JCL) to submit jobs to JES2 or JES3 (Job Entry System), which invoke, prioritize, monitor and manage the customer programs and utilities to access and update large datasets and databases using sort and other specialized utilities. Batch environment usually runs under the control of a job scheduler such as CA-7/CA-11.

JCL defines a series of job steps - a sequence of programs and utilities together with their required input and output files, and providing exception handling. Automated parsing and translation of JCL jobs to execute on UNIX/Linux or Windows environment enables the overall structure of the job to remain the same, including job steps, classes, and exception handling. Standard shell processing is supplemented with required utilities, such as SyncSort and support for Generation Data Group (GDG) files. REXX/CLIST/PROC scripting environments on the mainframe are similarly converted to ksh or other scripting languages.

A typical target rehosting architecture for batch includes a scheduler to control and trigger batch jobs, scripting framework to support individual job scripts, and an application server execution framework for the batch COBOL or C programs. Unlike other solutions that run these programs directly as OS processes without the benefit of application server middleware, Oracle recommends using container-based middleware to provide higher reliability, availability, and monitoring to the batch programs.

23


Integration with Oracle Scheduler or other job schedulers running in UNIX/Linux or Windows provides a rich set of calendar and event-based scheduling capabilities as well as dependency management similar to mainframe schedulers. In some cases, reporting done via batch jobs can be replaced using standard reporting packages, such as Oracle BI Publisher.

The target batch programs invoked by the scripts can run directly as OS processes, or if mainframe-class management and monitoring similar to JES2 or JES3 environment is a requirement, these programs can run as services under Tuxedo, benefiting from the health monitoring, fail-over, load balancing, and other application server-like features.

Files and Databases

Mainframes offer a wide range of file formats and databases, both prerelational and relational. Automated migration and rehosting approaches used to migrate them into a relational database or an equivalent filesystem depend on the type of the source data, and how it is used by the applications.

IBM DB2 is a predominant relational database on IBM mainframes. When migrating to Oracle Database, the migration approach is highly automated and resolves any discrepancies between the two RDBMS in terms of variable formats as well as error codes returned to applications, so as to maintain application behavior unchanged, including any stored procedures.

IMS/DB (aka DL/1) is a popular hierarchical database for older applications. Creating appropriate relational data schema for this data requires an understanding of the application access patterns so as to optimize the schema for best performance based on the most frequent access paths. To minimize code impact, a translation layer can be used at run-time to support IMS DB style data access from the application and map it to appropriate SQL calls. This allows the applications to interface with the segments now translated as ORACLE tables without impacting application code and maintenance.

VSAM files are used for keyed-sequential data access and can be readily migrated to ISAM files or to Oracle Database tables wherever transactional integrity is required (e.g., XA features). Some customers also choose to migrate VSAM files to Oracle DB to provide accessibility from other distributed applications, or to simplify the re-engineering required to extend certain data fields, merge multiple data sources, etc.

Other popular mainframe databases include CA-IDMS and Datacom, as well as Software AG’s Adabas. These databases can also be migrated to Oracle Database with a variety of automated migration tools and methodologies provided by Oracle Modernization Alliance (OMA) partners.

Meeting Performance and Other QoS Requirements

The mainframe’s performance, reliability, scalability, manageability, and other Quality of Service attributes have earned it pre-eminence for business-critical applications. How well do rehosting

24


solutions measure up against these characteristics? Earlier solutions based on IBM CICS emulators derived from development tools often did not measure up to the demands of mainframe workloads since they were never intended for true production environment and have not been exposed to large-scale applications. As the result, they have only been used for rehosting small systems.

Oracle Tuxedo was built to scale from the ground up to support high performance telecommunications operations. It has the distinction of being the only non-mainframe TP solution recognized for its mainframe-like performance, reliability, and Quality of Service characteristics. Most large enterprise customers requiring such capabilities in distributed systems have traditionally relied on Tuxedo. Consistently rated by IDC and Gartner as the #1 market share leader and predominant in non-mainframe OLTP applications, it has also become the preferred COBOL/C application platform and transaction engine for rehosted mainframe applications requiring high performance and/or mission-critical availability and reliability.

Reasons for the broad recognition of Tuxedo as the only mainframe-class application platform and transaction engine for distributed systems are based on mainframe-class performance, scalability, reliability, availability, and other Quality of Service attributes proven in multiple customer deployments. The following table highlights some of the key Tuxedo QoS attributes.

ORACLE TUXEDO MEETS MAINFRAME QUALITY OF SERVICE NEEDS

RELIABILITY AVAILABILITY

• Guaranteed messaging and transactional integrity

• Hardened over 25 years of use in the world’s largest

transaction applications

• Transaction integrity across systems and domains

through a two phase commit (XA) for all resources,

e.g., databases, queues, etc.

• Proven in mainframe-to-mainframe gateways

• No single point of failure, 99.999% or greater uptime with

N+1/N+2 clusters similar to IBM CICSPlex

• Application services upgradeable in operation

• Self-monitoring, automated fail-over, data-driven routing for

ultra high availability

• Centralized monitoring and management with clustered

Tux domains; automated, lights-out operations

WORKLOAD MANAGEMENT PERFORMANCE AND SCALABILITY

• Resource management and prioritization across

Tuxedo services

• Dynamic load balancing across domains based on load

conditions

• Data-driven routing enables horizontally distributed

database grids and differentiated quality of service

• End-to-end monitoring of Tuxedo system and

application services, enables SLA enforcement

• Virtualization support enables spawning of Tuxedo

servers on demand

• Parallel processing to maximize resource utilization with

low latency to maintain sub-second response at any load

• Horizontal and vertical scaling of h/w resources yields

linear performance increases

• Request multiplexing (synchronous and asynchronous)

maximizes CPU utilization

• Proven in credit card authorizations at over 13.5K tps, and

in telco billing at over 56K tps.

• Middleware of choice in HP, Fujitsu, Sun, IBM, and NEC

TPC-C benchmarks

25


As it delivers mainframe-like performance, reliability, and scalability in most demanding environments, Tuxedo is used by some of the largest TP applications worldwide to deliver over tens of thousands of transactions per second in real-world production applications, such as funds transfer, credit card authorizations, mobile billing, and reservations systems of major transportation vendors. It’s no surprise that most of the large rehosted mainframe applications in the 500 to 10,000-plus MIPS range are running on Tuxedo as well.

In larger deployments, Oracle Real Application Clusters (Oracle RAC (link)) are used to support the deployment of a single database across a cluster of servers—providing unbeatable fault tolerance, performance and scalability with no application changes. Oracle RAC supports usage of multiple individual systems as one clustered, virtual database server. It provides transparent synchronization of read and write accesses to databases shared by all nodes in the cluster, dynamic distribution of database workload, and transparent protection against systems failures.

As the further example of the benefit from integrated rehosting architecture using Oracle DB, RAC, and Tuxedo is Fast Application Notification (FAN) feature. FAN provides integration between the RAC database and Tuxedo. It allows Tuxedo to be aware of the current configuration of the cluster at any given time so that application connections are only made to instances that are currently able to respond to the application requests. The Oracle RAC HA framework posts a FAN event immediately when a state change occurs within the cluster. For example, for down events, Tuxedo can initiate transaction recovery.

In addition to delivering innovative and proven functionality in its own products, Oracle works closely with leading open-systems platform vendors to ensure that our products takes full advantage of their highly scalable systems with massive processor, I/O, and memory scalability, a complete resource partitioning continuum, granular workload management, and sophisticated operating system and system management capabilities. Combining these systems with proven Oracle Database’s RAC and built-in grid capabilities of Oracle Tuxedo and WebLogic Server, enables robust clustering with high availability, fast failover, dynamic load balancing, and unlimited scalability at much lower TCO than the mainframe. Oracle customers running mission-critical applications in the open-systems environment experience Quality of Service as high as, and in many scenarios better than, the same applications provided on the mainframe. Customer benchmarks comparing rehosted and original applications have repeatedly demonstrated greater transaction throughput on Tuxedo running on leading UNIX systems compared to the original application running on CICS or similar environments. In recent customer benchmarks of eXtreme Transaction Processing (XTP) applications comparing Tuxedo performance against the mainframe, Tuxedo has pushed the envelope beyond 100,000 transactions per second with application transactions that include computation and database I/O.

Phased Migration and Mainframe Integration

Some mainframe migrations are partial—to free up some needed mainframe capacity for other applications and avoid an expensive upgrade—and many full migrations are done in multiple

26


phases. In both cases, integration with remaining mainframe applications and mainframe-resident data is a critical consideration. Tuxedo Mainframe Adapters (TMA) provide this capability in mainframe rehosting projects as well as native Tuxedo applications when Tuxedo is used to run distributed services and co-ordinate access to mainframe applications and data for multiple front-end applications. TMA is available for TCP/IP, SNA, and OSI/TP networks (the latter used with Unisys mainframes) to deliver high-performance, bi-directional interoperability between applications running on Oracle Tuxedo and mainframe TP platforms, such IBM CICS or IMS TM .

CICS uses a set of InterSystem Communications (ISC) protocols for distributed transaction execution across multiple CICS regions. Tuxedo supports CICS ISC and supplies equivalent capabilities:

• Dynamic transaction routing, data-driven or based on load management policies

• Asynchronous processing to allow transaction execution to be started asynchronously from an invoking transaction leveraging in-memory and persistent queuing functions

• CICS Distributed Program Link (DPL)/Distributed Transaction Processing (DTP) functions provided by TMA, which supports transparent, bi-directional integration and global transaction coordination between mainframe CICS and IMS/TM applications and rehosted application on Tuxedo. This allows mainframe transactions to view Tuxedo as a remote CICS region virtually connected via APPC/LU6.2.

• Event-driven services infrastructure supporting a Publish/Subscribe model

• DOMAINS functionality providing full bi-directional connectivity and programming model across multiple Tuxedo application domains (similar to CICS regions)

Tuxedo Mainframe Adaptors provide bi-directional connections with full buffer mapping and propagate transaction context, including user id for security. Support for CICS DPL (CICS EXEC Link) and DTP (CICS EXEC verbs for LU6.2/APPC commands) facilities makes Tuxedo domains appear as another CICS region. This integration is provided over TCP/IP stack, allowing TCP/IP network connections to the mainframe while locally using SNA LU6.2 to connect directly from Tuxedo’s Communications Resource Manager (CRM) to CICS or IMS TM. Additionally, support for SNA connections allows you to use TMA without installing any new components on the mainframe.

Project Implementation

While each project is structured to meet specific customer requirements, a typical approach is often used as a baseline from which the actual project implementation plan is derived. The baseline structure comprises four main phases depicted in the table below. The first two phases represent an investment of effort and resources, while the following phases generate the actual payback.

27


TYPICAL STAGES OF A REHOSTING PROJECT

STAGE OBJECTIVE DELIVERABLES

Preliminary Questionnaire

Outline general requirements and

constraints, define target

architecture options

Preliminary Assessment

• Requirement summary

• Scope of work

• Project strategy and risk areas

• Target architecture options

• General project planning

Assessment Study

Evaluate possible options,

applications, and define target

solution and planning with budget

Statement of Work

• Refined architecture

• Refined ROI evaluation

• Risk mitigation action plan

• Refined project plan

• Project budget, conditions

• Pilot project detailed plan

Pilot or Proof of Concept

Validate target solution and

confirm budget and planning

Proof of Concept or Pilot Project

• Migrated pilot programs and data

• Regression test report

• Hardware and software platform benchmark results

• Final recommendations and refinements to the project SoW

Implementation Deliver rehosted solution and

support transition

Rehosting Completion

• Operational target infrastructure

• Operational migrated applications and data

• Renovated operations environment

• SOA-ready architecture

• Trained developers and operations staff

Further Modernization and SOA Enabling Rehosted Assets

To leverage and extend the value inherent in mainframe applications, rehosting is often followed by service enablement for integration into an SOA framework. Integrating rehosted applications into the SOA framework provides key benefits:

• Improves productivity, agility and speed for both Business and IT

• Allows IT to deliver services faster and align closer with business

28


• Allows the business to respond quicker and deliver optimal user experience

• Masks the underlying technical complexity of the IT environment.

When business value of rehosted applications requires integration into an enterprise SOA, the approach must maintain and extend the applications’ QoS attributes. Key considerations for integrating rehosted applications into an SOA include:

• Achieving appropriate services granularity

• Understanding requirements for transactional integrity and reliability

• Ensuring end-to-end messaging security, credential propagation, and other security policies

• Defining expected response time, throughput, and scalability and enabling SLA management

• Providing support for heterogeneous client connectivity

• Leveraging service orchestration and BPM integration

• Enabling services lifecycle governance through a set of clear policies and supporting processes

Rehosted applications running as Tuxedo services can take advantage of SOA integration and enablement capabilities provided by Oracle products shown in the table below.

SOA ENABLEMENT AND GOVERNANCE SOLUTIONS FOR ORACLE TUXEDO

REQUIREMENTS ORACLE

PRODUCT

KEY FEATURES BENEFITS

Web Services

gateway for

point-to-point

integration and/or

Service

Composition

Architecture

(SCA) bindings

Oracle SALT • Inbound and Outbound WS:

SOAP/HTTP(S)

• Extensible XML Data Mapping

• WS-Security

• WS-Addressing, WS-RM, MTOM

• WSDL Creation and IDDI Publishing

• SCA Container and Bindings

• Open standards-based integration

• Faster time-to-market

• Lower integration cost

• Avoids the need to rewrite in Java or

.Net

Multi-protocol

Service Bus for

orchestrated

integration

Oracle Service Bus (OSB)

• Adaptive, heterogeneous messaging

for SOAP, EJB/RMI, JMS, Tuxedo,

IBM MQ, SAP, SWIFT, FTP, etc.

• Extensible message brokering

• Dynamic routing with multiple

transports and transformations

• Transactional support (XA)

• Embedded management with

monitoring and reporting

• Pluggable Security

• Faster deployment and simpler

management of shared services

across the SOA

• Rapid transformations

• SLA-based monitoring

• Service life cycle management and

governance

29


Enterprise

Repository for

service discovery

and lifecycle

governance

Oracle Enterprise Repository

• Usage Tracking

• Policy Management

• Dependency Analysis

• Metrics & Analytics

• Interactive Navigation

• Visibility, traceability of service

assets, their lifecycle governance

• SOA services governance

• Greater reuse of rehosted

applications

Business

Process

Management for

orchestrating

system and

human workflows

Oracle BPM and BPA Suites

• Native BPEL Support

• Comprehensive process modeling,

simulation, and dynamic publishing

• Visual monitoring and auditing

• In-flight administration

• Broad connectivity: JMS, email, JCA,

web services, and packaged adapters

• XSLT and XQUERY Transformations

• Rich human workflow services

• Robust BPEL engine for scalability

• Maintains context of long-running

flows in DB for clustering and HA

• Enables IT and line-of-business

collaboration through process

modeling and event simulation

• Eliminates the boundaries between

applications, systems, and people

• Gives visibility and control to

business process owners through

alerts, KPIs, and dashboards

• Speeds delivery of new solutions

with out-of-the-box features

• Immediate ROI with advanced “what

if” analysis for optimization and

automation of key processes

The migrated application components running in Oracle Tuxedo have preserved the business logic and UI. But now we can also expose suitable service interfaces in legacy components via standard WSDLs and provide robust bi-directional web services gateway capabilities. These can be used to integrate with packaged or custom applications, including Oracle Applications or 3rd party ERP systems. Rehosted COBOL services can also leverage Oracle Service Bus with built-in Tuxedo Transport for heterogeneous messaging. Note also that because Oracle SALT and OSB are bidirectional, rehosted components in Tuxedo can call outbound web service. This can be done by simply mapping a transaction name used to call locally to the name of a web service and leveraging SALT’s Web Services gateway or OSB’s Tuxedo Transport for protocol and XML conversions. The web services and ESB end points can all be imported into Oracle Enterprise Repository to start providing visibility, tracking of service dependencies, and other benefits of service lifecycle governance. All this can be done without rewriting any COBOL code.

The capabilities provided to rehosted applications through Oracle SALT, Oracle Service Bus, and Oracle Enterprise Repository can be leveraged in phases or as part of a single integrated initiative. Extending the rehosted applications through SALT for Web Services integration or through Oracle Service Bus for heterogeneous service messaging is a simple initial step on the road to SOA enablement. Customers use SALT for its complete open-standards web services capabilities that easily integrate with any web services environment. For customers who need integration with heterogeneous messaging beyond SOAP/HTTP, require global transaction coordination with other XA-enabled components, and value strong transformation,

30


orchestration, and management capabilities provided by Oracle Service Bus, this approach provides a powerful way to extend and integrate rehosted applications.

Following the initial integration steps or in parallel with them, customers can leverage Oracle Enterprise Repository to provide a single meta-data repository populated with services information for governing the discovery, deployment, and full life cycle of these rehosted application services, enabling greater leverage of these key assets throughout the enterprise.

Figure 7. Reusing legacy services in orchestrated business processes via Oracle BPM

A further step in SOA-enabling and modernization of the legacy applications is to begin leveraging rehosted services in BPM-driven dynamic business processes. Reusing rehosted legacy application services via BPM, such as Oracle BPEL Process Manager, unlocks the siloed logic and puts it to use as a strategic enterprise asset. The objective is to lift these “high touch”, frequently changing aspects of the application above the application code and use declarative models and tools that make it easy to adapt and quickly respond to business needs.

The diagram below illustrates how rehosted legacy services can be leveraged by a BPM framework. BPM design-time tools can easily discover the available services exposed via Oracle SALT or Oracle Service Bus using services meta-data published in Oracle Enterprise Repository. Connecting to the service via Web service interface (provided by SALT), or an ESB business service (provided by Oracle Service Bus), handles run-time binding and access.

31


Further Rearchitecture of Rehosted Applications

Rehosting application logic intact helps to preserve the investment in the code, and simplifies ongoing maintainability leveraging current resources. But what if we need to add new capabilities or extend existing logic significantly in some areas? You have a choice. You can continue to leverage the COBOL skills or C/C++ in Tuxedo, or leverage J2EE app server like Oracle WebLogic and write new components in Java as EJBs or a Plain Old Java Objects (POJOs.) It’s all transparent and seamless. From Tuxedo perspective any application components in a local or remote Tuxedo domain or WebLogic server can be easily accessed and the bi-directional connectivity between COBOL and Java provided by Tuxedo and WebLogic is the best in class.

Most recent Oracle Tuxedo release 10gR3 extends these options by adding support for Services Composition Architecture (SCA) components in Tuxedo. SCA is a next generation of SOA technology—going beyond Web Services to the realm of multiple protocols and execution environments, language independent orchestration of services using more of an RPC style call interface. Oracle SALT 10gR3 enables Tuxedo to host SCA containers, and provide client and server-side bindings for ATMI, SOAP, and Java. The inherent characteristics of SCA enable programmers to focus on business logic—reusing existing components is made easier and rewiring them is highly automated. Although components may run in multiple container types and use different protocols, these are compatible and accessible through SCA composition tools. Support of SCA programming model on top of Tuxedo renders the ability to more effectively manage service lifecycle, supporting systematic reuse of existing services (including rehosted COBOL components) in SCA-based composite applications, as well as runtime discovery of service signature.

In an SOA-like environment of Tuxedo application services, isolating individual components and replacing them with calls to rearchitected Java services or SCA components is transparent. This robust integration between rehosted and rearchitected components completes the value circle to maximize the benefits from leveraging existing application assets.

Conclusion and Customer Examples

Making the move off the mainframe via mainframe rehosting has enabled numerous customers to lower costs, and helped IT deliver greater business impact due to flexibility of the open systems environment and greater agility of the application services rehosted to an extensible, SOA-ready platform.

A key factor in a successful rehosting project is the application platform that combines proven ability to run mainframe applications without requiring invasive changes with the required performance, availability, and reliability, while making it simpler to SOA-enable and/or rearchitect selected components as needed.

32


Oracle’s rehosting solution for mainframe modernization provides a proven, low risk, and highly efficient way for enterprises to migrate critical online and batch applications off the mainframe to lower-cost open systems environment powered by Oracle Database and Oracle Tuxedo. The benefits of using this solution are summarized by contrasting the before and after IT environment in the table below.

CONTRASTING IT ENVIRONMENT BEFORE AND AFTER MAINFRAME REHOSTING

BEFORE REHOSTING AFTER REHOSTING

• 70-80 percent of IT budget spent on IBM and other

mainframe costs and resources

• Proprietary and closed environment with shrinking skills

base

• Costly and brittle integration of new applications, big

drag on time-to-market

• Expensive to adapt to the needs of the business

• High level of backlog for mainframe application

changes

• Ongoing annual cost savings of 50-80 percent compared to

premigration maintenance costs

• Reduced or eliminated IBM and CA dependence and lock-

in

• Increased agility—core enterprise assets are SOA-enabled

for a standards-based and composition-driven assembly

model

• Faster application changes with enhancements delivered

by Java/.Net components transparently integrated with

legacy COBOL code

• Funds and resources available for new initiatives, reducing

backlog

This approach can accelerate the cost savings, shorten time to meaningful ROI, and substantially reduce legacy modernization risks. It balances the benefits of the powerful open systems-based platform and software stack with the low risk, non-invasive migration of the invaluable application logic and data. For many customers this represents an optimal first step in their modernization roadmap—an approach that quickly reduces costs and generates positive returns, and puts them on a solid financial and technological footing for further modernization investments.

Substantial cost reductions have been achieved by many banks, insurers, manufacturing and service companies, as well as government organizations. A number of these customers have taken advantage of the flexibility of the new environment and rapid reuse of the rehosted application components to extend their applications using modern development paradigms, rich SOA integration options, and ability to easily integrate rehosted services with packaged applications.

“This target architecture is CNAM’s first concrete step in component-based development. Functions are now reusable.

Development and Maintenance workload is reduced. Improved responsiveness and IT performance resulted in faster

implementation of new features.”

Mme. Maryvonne Cronier, CIO, CNAM-TS

33


While the TCO reductions depend on many factors, such as the scope of the migration, complexity and scale of the rehosted applications, preferred target environment, etc. we can provide examples from past projects presented by customers at BEA World and Oracle OpenWorld.

MAINFRAME REHOSTING CUSTOMER EXAMPLES

PROJECT DESCRIPTION RESULTS

PG&E: A US gas and electric utility serving over 13M customers in Northern and Central California

PG&E has migrated CICS/COBOL/DB2-based Customer

Care and Billing application suite to Oracle Tuxedo,

WebLogic, and Oracle Database/RAC cluster. The suite

provides applications handling:

• Measure, bill, collect processes to support meter

reading

• Meter data management and rendering of bills,

• Customer records, credit and payment processing

• Contact Center and Front Counter Support for customer

transactions such as start/stop and account status

• Scheduling and dispatch of customer field orders

The migration has enabled PG&E to:

• Scale the application dramatically to support PG&E

SmartMeter™ rollout of 10.3 million new electric and gas

meters that electronically report consumption daily or

hourly. The transaction volume has scaled to 1 Billion

select/fetch and 100 Million inserts/updates/deletes daily

• Gain performance improvements - 28% gain in batch

processing and 25% improvement in online processing.

• Maintain their 7x24x365 uptime requirements

• Generate $5M in annual savings by reducing mainframe

footprint significantly (from four mainframes with 13

processor down to single mainframe with one processor)

CNAM-TS: A Social Security organization handling health insurance claim processing for 50M members in France

CNAM-TS has migrated its 12,000 MIPS worth of

mainframe applications responsible for processing 1.1B

claims a year and issuing €110 Billion in reimbursement

payments annually. They undertook IT consolidation and

mainframe rehosting to reduce IT costs and increase

agility in responding to regulatory changes in health care.

CNAM-TS rehosting project migrated 3,500 COBOL

programs, 4,600 JCL jobs, and 270TB of data from two

mainframe architectures (IBM and BULL) to a single

architecture on open systems environment powered by

Oracle Tuxedo and Oracle Database.

CNAM-TS benefits after mainframe rehosting:

• IT consolidation and simplification by rehosting

mainframes to a more flexible open systems architecture

• Organizational transformation, which created the

foundation for CNAM’s move to component-based

development and enabled Java extensions to COBOL

applications, contributing to IT skill evolution

• Reduction of daily processing by 8 hrs., which enabled

introduction of smartcard services on the same systems

• Improved reliability/availability by deploying across

multiple sites with horizontally distributed, replicated DB

and load balancing via Tuxedo data-driven request routing

• Cost reduction from $65M a year down to $10M a year for

85% cost savings to deliver the same capabilities on an

open systems infrastructure without the mainframe

34

Mainframe Rehosting Leveraging Oracle Tuxedo: Accelerating Cost Reduction and Application Modernization June 2008, Updated March 2009 Author: Mark Rakhmilevich Oracle Corporation World Headquarters 500 Oracle Parkway Redwood Shores, CA 94065 U.S.A. Worldwide Inquiries: Phone: +1.650.506.7000 Fax: +1.650.506.7200 oracle.com

Copyright © 2009, 2008 Oracle and/or its affiliates. All rights reserved. This document is provided for information purposes only and the contents hereof are subject to change without notice. This document is not warranted to be error-free, nor subject to any other warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or fitness for a particular purpose. We specifically disclaim any liability with respect to this document and no contractual obligations are formed either directly or indirectly by this document. This document may not be reproduced or transmitted in any form or by any means, electronic or mechanical, for any purpose, without our prior written permission.

Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Other names may be trademarks of their respective owners.

0109

mainframe re-hosting with oracle tuxedo: accelerating cost

Documents