For beiT SOA is like the plattform idea of the automotive branch.
SOA away from the abstract theory
Often as arguments for the introduction of service oriented architectures (SOA) targets like lifecycle-governance and/or process-orchestration are used but in real world these targets are not achievable in near future due to very often involved legacy systems. Based on the evolutionary approach of SOA existing legacy solutions can be transformed and migrated to services step by step. According to that beiT recommends introduction of SOA based on initially small projects as essential success factor. SOA is definitely no panacea to master architecture issues. The difference to former approaches (like CORBA, COM/DCOM, EJB) can be find in the availability of open standards and big support of all relevant IT vendors. SOA enables therefore a simpilfication and flexibilisation of mostly numerously software interfaces through the elimination of redundant functionality without getting dependent of vendors.
beiT tipp: Start with collecting experience with small SOA integration projects and achieve quick ROIs - business process analysis of the whole enterprise then afterwards.