In this article, we examine promising new mainframe migration tools by BluePhoenix Solutions and LegacyJ Corp. that join the tools of suppliers such as MicroFocus Ltd. Technically, they demonstrate the real-world ability to carry out migration in specific cases but also the ability to upgrade mainframe applications in place. In other words, today's migration tools do not absolve the IT strategist from careful consideration of the merits of the mainframe platform, and they do give the IT strategist new abilities to improve mainframe applications, whether the application is migrated or not.
Mainframe migration toolmakers
Israel-based BluePhoenix offers a variety of mainframe migration tools and services. Its offerings can be used for legacy-application discovery, application portfolio management, migration of applications from one database/file system/platform to another, rewriting applications in a new programming language and infusing existing applications with new technologies, such as Web services.
Key tools include:
BluePhoenix also offers tools for upgrading and standardizing code and data (e.g., implementing euro data according to euro financial standards). The most recent BluePhoenix release adds service oriented architecture (SOA) enablement capabilities.
The company's credibility is enhanced by two key bits of technology:
1. A year 2000 tool. Year 2000 tools have proved conclusively that even the most business-critical mainframe applications can be changed without harm to the business, and these tools embody plenty of real-world code-changing experience.
2. A re-engineering tool based on Seer Information Technology Inc.'s outstanding Seer HPS development tool. This tool captures an abstraction of a mainframe application and then regenerates it to Java code. While much of the speed-to-market advantages of this tool are lost when it is used only for the purpose of re-engineering, its wealth of knowledge about mainframe environments enables it to re-engineer an unusually large set of mainframe applications in a highly automated and flexible way.
BluePhoenix has begun to attract and carry out important, although not mammoth, migration projects -- a multimillion-dollar migration to Windows/Unix platforms for DesJardins General Insurance Group in Canada. The company remains small and is growing moderately, with $59 million in revenues in 2005. About 50% of BluePhoenix's market presently involves migration from the mainframe. However, nothing in BluePhoenix's technology prevents it from providing equally effective services that will upgrade mainframe applications in place.
Another player in the mainframe migration market is Los Gatos, Calif.-based LegacyJ. LegacyJ's target platform is a Java virtual machine (JVM), which can run on most platforms, including mainframe Linux.
LegacyJ offers an Eclipse-based COBOL compiler that compiles all major flavors of COBOL into Java byte code, deployable via the JVM to any hardware platform. The resulting COBOL and Java programs can directly call each other. The migrated application retains both the source code (as COBOL) and the user interface of the original application, so the developer or end user requires little to no retraining.
Accompanying LegacyJ's compiler is a Java-based CICS platform, called LegacyJ Transaction Platform, to support migrated CICS-based applications, and data bridges to allow migration to a relational database. Thus, LegacyJ provides z/OS-equivalent services on the target platform.
Recent successes in migration and in boosting the technology of the mainframe in such areas as Web services suggest that there is more technology improvement in IBM's mainframe-installed base than at any time in recent memory -- and about time, too.
Infostructure Associates continues to argue that upgrade in place is preferable to migration to another platform in the majority of cases because upgrade in place involves less risk, especially for large enterprises. and allows an incremental approach rather than all-or-nothing migration. At the same time, users should consider the long-term effects of these trends. At some point, with most mainframe and nonmainframe applications able to be ported to another machine or upgraded in place, IT strategists will no longer be able to hide behind "these applications can't be moved" or "the mainframe is dying" -- neither is likely to be true. Instead, users will need to make migrate vs. upgrade decisions purely on the "fit" of the platform with the user's needs. The mainframe will fit for some needs; another platform will fit for others.
In other words, users should accept two take-away messages from recent improvements in mainframe migration tools -- which in most cases are also upgrade-in-place tools.
1. It is time to reconsider the "leave alone" strategy for mainframe business-critical applications.
2. Both migration and upgrade in place are increasingly feasible alternatives.
About the author:Kernochan is president of Infostructure Associates.