Today there is no best practise method available to effectively estimate the maintenance costs of historically grown large-scale software landscapes. Most cost estimation models are either not generalizable due to highly specialized scenarios or too abstract to be implemented in practice. In this paper we introduce a multi-level approach to create transparency, estimate costs realistically based on current spending and establish a method for sustainable cost-control. At the heart of our approach is the deduction of meaningful indicators for estimating current and future maintenance efforts. We present the first version of a statistical cost estimation model being implemented at Deutsche Post MAIL as a baseline for contract negotiations with providers.

No abstract