Automated Software Modernization Costing Model

With automated software modernization you tend to get what you pay for. Free Wizards and do-it-yourself conversion tools don't live up to expectations. Canned tools leave as much as 10% to 40% of the most difficult code unconverted. Any code left incorrectly transformed becomes a manual task requiring human intervention as a time and materials activity, either by the in-house programming staff or an external company.

The Gartner Group estimates that a well trained programmer can modernize at most 160 lines of code per day manually. Programmer productivity fall offs sharply when code for either the legacy source or target language is more complex. The table to the right estimates software modernization project costs for easy, medium and hard language pairs at different levels of automation for a 1 Million Line system. For an easy conversion the Gartner estimate of 160 lines of code per day is used. For medium and hard conversions more realistic manual rates of 40 lines of code and 20 lines of per day are used.

For a 1 million line system a transformation tool or service capable of only a 60 percent level of automation would require 2,500 man-days of manual intervention for an easy language pair using the best case 160 LOC per day programmer productivity ratio. The cost to finish just the code conversion (exclusive of other project costs, such as system architecture, testing, deployment, program management) modernization would vary from $1,000,000 to $3,500,000 using labor rates from a high of $175 per hour (domestic consulting) to a low of $50 per hour (off shore) labor rates. For medium and difficult language conversion the cost of 60% partial solution ranges from a low of $4 Million to a High of $28 Million for the 1 Million line system.

By comparison a transformation service delivering 99 percent automation would leave only 10,000 lines of code untransformed and requiring human intervention.