The application systems that have been developed for mainframe computers over the past 30 years are built around the traditional Legacy System architecture of a batch COBOL system that is supported through an OLTP, which is almost always CICS. In fact the predominance of such Legacy Applications cannot be overstated. Even today, nearly 75% of existing business software applications follow this model.

The business requirements for IT organizations have of course, changed through the years as did available technology. Remaining essentially the same over time, though, legacy applications are still the core of most organizations, however the need to integrate new platforms with corporate applications and data, develop internet strategies and develop new architectures to support future needs have all combined to further complicate today’s environment.

In addition to these more technical issues, there are other considerations that must be recognized as being equally important.

The ability to attract and maintain a qualified staff to handle an increasing workload on legacy systems has reached the point of concern in many metropolitan markets. Some organizations have found it necessary to train their own programmers due to such critical shortages of available skill sets. This trend is forecasted to increase in the future as educational institutions from technical schools to universities are no longer including COBOL and CICS in their curriculum.

Financial issues including the need to recoup Y2K costs have also placed additional burdens on IT management. Many organizations spent up to an entire 18-month budget on Y2K alone. While Y2K remediation is certainly understood to have been a requirement, there have been lasting effects more so financially than in any other way.

An additional ripple effect of Y2K was the application change ‘freezes’ that many organizations imposed upon their systems departments preceding and following January 1, 2000. While these steps were a necessary part of Y2K preparedness in ensuring that corporate systems were stable and in healthy conditions. The net effect has been another obstacle to be overcome in the form of lost time.

Today, in an effort to move legacy applications forward, there are several approaches being used in application transformation and migration.

Application Transformation, as it is defined today, is essentially an application development effort designed to convert COBOL, and in some cases CICS, source code to another language / platform / architecture. Generally, these conversions concentrate on moving applications to Java via ‘rules mining’ – a process whereby the logic of a COBOL program is extracted and used to create an Enterprise Java Bean. Transformations using these techniques do result in platform independent, web enabled applications, however the cost and time required for a complete working solution is extremely high and for most organizations is not a viable alternative. Current prices for turnkey COBOL to Java conversions range up to $25.00 per line of COBOL source. Disregarding the enormous cost, there is still the concern that while the application has been re-deployed and is a platform independent, it is still essentially the same application and generally offers no immediate advantage in terms of functionality.

Direct migration, the task of moving an existing application to another hardware platform, such as moving a COBOL CICS based application from a mainframe to a UNIX platform has historically involved making significant source code changes. Dealing with the idiosyncrasies of the various levels of COBOL, platform dependent compilers as well as emulators for BMS (IBM’s 3270 based display layer, a.k.a. Basic Mapping Support) as well data access requirements has generally made moving from one platform to another difficult.

Additionally, to date, there has not been any non-mainframe based CICS software that has any benefits to offer an organization. Primarily PC and UNIX based CICS software has been intended as a development environment for mainframe programmers. In other words, the three primary goals of application migration: 1) easing application integration 2) providing a production capable environment and, 3) positioning applications within an architecture capable of easily supporting web interfaces and other future needs, have not been addressed by cross platform OLTP developers.

At RMS, having worked with small businesses up to Fortune 100 corporations we have had first hand experience witnessing the issues we’ve discussed here. As a result of these experiences and our technical expertise in applications development, mainframe systems/programming and PC’s and networking, we have been able to bring together a unique collection of knowledge and talents, without which a system such as Eden Server would not be possible.

Starting from the concept phase, the goal of developing the Eden Server system has been to provide organizations with an alternative approach to legacy systems management and support. The result is a Windows based OLTP system that, by bridging the gaps between available technologies, has created an entire new set of alternatives for managers of legacy applications.

The Eden Server system is a Windows application that serves as a TCP/IP based OLTP to support COBOL CICS legacy applications. The features of Eden Server are:

• Eden Server and Eden Client are native Windows applications that run on Windows 98 through Windows 2000 and require no special operating system or network configurations.
• `No source code changes required’ support for OSVS through VSC4 on Windows platforms using a NetExpress 3.0 or higher COBOL compiler in conjunction with RMS’s pre-processor for CICS.
• Full support for mainframe file formats including Indexed, Relative and Sequential. All files built to industry standard Micro Focus formats. Large file sizes up to 8 MTb (MegaTerabyte) are supported. Mainframe print format files are also supported with built in utilities used to format reports for network based printers.
• Full support of standard ‘BMS’ mapping features via RMS’s BMS Macro import utility which converts BMS assembler macro statements into native files for LAN use, or to Java scripts for access via a browser across the internet.
• Complete GUI management console, including a true system ‘console’ and over 100 real time monitors covering every vital system measurement required to manage the communications network, system resources and a built in connection firewall.
• Support for up to 1000 client connections via high performance overlapping network I/O routines developed by RMS. Because Eden Server and Eden Client use RMS developed networking modules, there are no special ‘per user’ licensing requirements for the server machine other than a one user version of Windows NT or Windows 2000.
• Support for running up to 5 concurrent Application Regions on one Eden Server machine.
• Point and click installation and management of Application Regions, including automatic building of most system information.
• All COBOL source and BMS maps compiled and linked to native DLL and exe formats.

As this brief feature list begins to show, Eden Server is built to handle the needs of even a very large organization. Because of pricing and on going cost saving potential, however, even the smallest organization will find that Eden Server is an extremely viable alternative to the status quo.

Technically, Eden Server provides organizations with a means to an end. By designing Eden Server as a three tiered application in and of itself, legacy applications are automatically positioned to take advantage of new technologies without the need for screen scrapers or other middle-ware. Tasks such as developing a new presentation layer to replace character BMS screens in languages such as Visual Basic are now possible and due to the common platform of Windows can even be done in a staged approach one screen at a time. By using any readily available web server and Java client software it is possible to immediately deploy legacy applications on the internet.

In addition to being full featured Eden Server exploits Windows NT and 2000 architecture with it’s own multi-threaded and multi-process framework so it is an extremely stable platform. Surveys of current Eden Server clients indicate it to be much faster than recent predecessors such as UNIKIX, MTS and CICS/NT. Unfortunately accurate performance comparisons are not available due to the lack of such information from other such systems. Eden Server, produces a full range of performance measuring figures, including number of transactions per day and average ‘terminal’ response time. On systems as small as 10Mbs Ethernet, 333Mhz single CPU servers, Eden Server typically processes 50,000+ transactions per 8 hour shift with average response times of 0.5 seconds; with an average server CPU load of 20 percent.