Since the late 1980s, we have seen many solutions hit the market. Below is an overview all the options we typically see. Generally, there are 5 Modernization/Migration Options:
- Fully Automated Modernization
- This option fits well for large, single or multi-language applications that are actively maintained, valueable and irreplacable. TSRI’s solution uses model-driven, AI-based, fully automated methods to achieve 100% automation in the modernization of a source application to a target language and architecture of your choice. Learn more about the advantages of the fully automated approach here.
- Semi-Automated Modernization
- Semi-Automated Modernization technologies an work, if the source application is small or very small and you have a specific reason to choose a less automated method. However, because of the manual input required to fill gaps in the transformation, the transformation risk with semi-automated modernization increases corresponding to the size of the application, because humans are in the loop touching the code and introducing errors. For example, transliterators (which translate code “line-by-line”) produce legacy-like target code that must be manually edited. Other semi-automated model-driven approaches require additional manual rewriting, often for large amounts of code at large expense and risk. This increases the chances of schedule overrun and increasing costs, and does not allow for automated refactoring, or flexible approaches that allow for mid-project changes.
- As a result, semi-automated solutions that can handle one or two languages are typically not good solutions for larger projects. Even an automation level of 95%, on a 1,000,000 line application, means that 50,000 lines of code or more must be manually written, at dramatic expense.
- Application Rehosting & Replatforming
- Rehosting is a good solution if the application is not actively maintained and only needs to move off its current hardware/platform. This solution retains the legacy language, and so development/maintenance continues to be difficult and costly, and exposes the customer to increased technical debt as fewer and fewer resources are available. Also, the significant license costs associated with Rehosting can be unattractive for customers. In fact, the effort to undertake a TSRI modernization is often equal or lower than 1-2 years worth of licensing cost.
- Custom Off-the-Shelf (COTS) Solutions
- COTS replacements are very common for replacing general applications that do not have many custom features. Usually the COTS applications look completely different, have different functionality and have license costs associated with them. This requires costly and lengthy retraining of the current users who are not familiar with the new system and rely on the vendor to “make-them-work” for the business scenarios. Again, license costs and vendorlock are other disadvantages, as is the ability to maintain and modify the application moving forward.
- Manually Rewriting Applications
- Definately the most costly, highest risk and longest approach is rewriting the code manually. This is an approach often used on applications that need massive forward development and required major changes in functionality. The independant research firm, the Standish Group, shows clearly that the high liklihood of “challenged” or “failed” projects, based on hundreds of modernizations/migrations/replacements, should be a major consideration for any manual redevelopment effort. Most of TSRI’s successful large modernizations follow a failed manual effort that expended millions or tens of millions of dollars before being scrapped. Request a copy of the Standish Group report here.
Learn more about modernization options from independent 3rd party research:
Standish Group Podcast on Modernization Options:
