The Legacy Systems Dilemma

Since legacy systems are systems that by definition have been inherited from the past, they can pose special challenges for an organization. This is especially the case as these systems reach end-of-life, when supporting the status quo becomes difficult. The pain of not doing something new increases as business benefits erode. The trouble is, the process is insidious. Like a garden slowly falling into neglect, it’s hard to see what’s happening on a day-to-day basis. The problem creeps up on you. Then, one day, you can suddenly be faced with the fact that your legacy systems are no longer affordable. And not just in terms of how big a slice of the budget they require, but because they divert funds from new initiatives essential to business competitiveness.

Rather than endure this insidious process, why not examine the status quo and develop an ongoing roadmap for change? Proactively analyzing some of the less-obvious effects of legacy systems can give you “fresh eyes” to see how badly legacy systems are affecting your company’s ability to do business. For example:

• Reliance on tribal knowledge Legacy systems must rely on a shrinking “tribe” of practitioners for operation and maintenance. The necessary skills, such as COBOL and CICS, are not being taught anymore. Ask, “How dependent on tribal knowledge is my organization today?”
• Lack of extendibility Obsolete development tools and interfaces can make legacy systems difficult to extend, integrate, and enhance. Today, systems touch all parts of the business: front office, back office, the web, remote users, etc. Legacy systems can’t “plug and play” to support cross-functional business processes. Ask, “How many cross-functional activities do my systems support?”
• Duplication of effort Sometimes more than one legacy system supports the same capability, or, conversely, many unique systems support custom projects. In addition, it is not unusual to have thousands of reports being generated when only a few are needed-and it is sometimes impossible to capture information in one place about several projects or cross-functional business processes. This lack of “business intelligence” makes running the business much more challenging. Ask, “How easy is it for decision-makers to get the information they need to support the business?”
• Regulatory non-compliance Legacy systems can be difficult to change and may not respond easily to new regulations like the Sarbanes-Oxley Act and HIPAA, which require companies to store and retrieve data across all types of media (email, voicemail, documents, digitized images, etc.). Ask, “Can you accept the risk of not complying with new regulations?”

Asking questions like these will help clarify the need for change. The next important step is to develop and execute a roadmap that will make change less painful and more successful.

Roadmap to Success

Moving away from a mainframe or similar legacy system need not be an “all-or-nothing” proposition. A migration roadmap for success will include key elements such as those highlighted below. As the enterprise begins to implement this type of roadmap, business benefits can increase and total-cost-of-ownership (TCO) can decrease. Of course, there are advantages and caveats for each of these roadmap elements. All can lower costs and enhance functionality.

Emulate

A low-risk approach is to emulate the existing applications or process on a more cost-effective platform.

Advantages

• Low risk, familiar function and process
• Requires almost no changes to code or data
• Provides a solution for applications operating in environments at full capacity
• Can be used as an enabler for large scale legacy transformation

Caveats

• Long term return on investment (ROI) and benefits can be less
• May not make the application more flexible

Package

The strategy for “packaging” is to match applications and workloads to dedicated, optimized servers.

Advantages

• Performance optimized platforms
• Increases ROI and streamlines processing
• Increases data availability
• Enhances application interoperability

Caveats

• Data quality and application incompatibilities can create problems
• More potential risk
• Requires understanding of architecture and application portfolio

Re-platform

The strategy here is to move to replace in-house systems with off-the-shelf, packaged applications.

Advantages

• Can help align both business and technology strategies
• Helps avoid complex application migrations
• Has the potential to reduce overall maintenance costs and increase functionality

Caveats

• Packaged applications can be expensive, and they are not tailored for a specific environment.
• Staff skills may need to be augmented.

Convert

The strategy for conversion is to preserve existing investments in code and data, and to extend and enhance legacy systems with modern technology. It is important that conversion results in the execution of “native” code in the new environment.

Advantages

• Better environment for ongoing maintenance and development
• Leverage modern, efficient and cost-effective technologies
• Preserves and enhances existing application investments

Caveats

• Assumes knowledge of business logic
• All data dependencies have to be understood.

Sheer “Horsepower” Is No Longer the Issue

Modern network operating systems allow you to add virtually unlimited processing power, similar in concept to Lego® building blocks. In fact, Windows .NET has evolved to be far more than “a bunch of networked PCs that will never replace the mainframe.” The Windows .NET Mainframe Server is:

• Data Center Certified for capacities up to 1000 MIPS.
• Designed for interoperability using Web Services standards
• A common foundation for back-office and front-office computing
• Highly available options with cluster and fault-tolerant technology
• High performance for transaction-intensive applications and batch operations.

Moreover, this is not “bleeding edge” technology for early-adopters. A case in point is the State of Washington Department of Licensing (WADOL), which has a mature set of applications running on a Unisys 2200 mainframe. These applications, developed over 30 years, comprise close to 1.5 million lines of COBOL code, accessing thousands of files, large databases (DMS and RDMS), and hundreds of user interface screens, and were controlled by over 50,000 lines of job control language (Unisys’ ECL and SSG).

The Department recently decided to migrate all the applications supporting the various licensing activities, including vehicle licensing and drivers licensing and other administrative services, to the .NET Framework using COBOL or Visual Basic by May 2005. The migration is well underway, and the Department expects to achieve the following goals:

• Cost savings of over $1 million per year in applications maintenance
• Significant productivity gains, with less dependence on outside contractors and improved responsiveness to legislative changes.
• Elimination of the need for specialty Unisys skills.
• Access to thousands of off-the-shelf products, providing greater power to users and efficiencies in the development process.
• Better systems integration, removing the need for awkward workaround processes.
• Improved information access through the tools and functions available from storing the data using SQL Server. Both the usability and quality of data are expected to improve.

A roadmap approach, combined with the existence of viable technology, allows any company to do the same today, to whatever degree makes sense for a particular environment.

An Ongoing Process

In summary, every successful system is destined to become tomorrow’s “legacy.” Legacy migration is therefore an ongoing process. Asking the right questions, making sure that change isn’t creeping up unawares, and having a good plan in place are they keys to successfully rolling over one legacy system to another. Change needs to be incremental and ongoing, so that the pain is spread out appropriately. Roadmap choices allow you to manage that pain, and to avoid some of the more insidious effects of legacy systems on business competitiveness.