How to make legacy point-to-point system flexible to comply with changing business requirements
October 20, 09:35
I’m going to share with you how we managed to switch legacy point-to-point enterprise system to concept of integration architecture. Being integrated a component of the system gets rid of firm dependencies of neighbor components, can be developed independently but keeps and even extends its ability to reach other components.
To get these advantages in running system we finally designed and implemented an integration component, which broke bad practice of changing code to maintain sequence of dependencies and which introduction allowed
- to model and to execute business processes describing dataflow among several components. Modification of a business process can be applied after application restart, without rebuild or redeploy of the component;
- to process interface changes by means of configuration, if connected components still contain enough data to communicate;
- to keep compatibility of the entire system if one of the components failed to be released, flexibly change components landscape and introduce new functional components as soon as they are developed or to retire old subsystems replacing them by set of business processes.
I will describe the stages we went through to archive the goal, explain why we had to create internal DSL working inside integration component and share with you a couple of unexpected features we could easily implement by means of configuration.
My address can be interesting for those who have to maintain and develop big legacy systems and finds this process painful.
Lead Analyst, T-Systems
St. Petersburg State Technical University degrees in specializations “Telecommunication systems and networks” and “Mathematical supply and administration of informational systems”. 10 years of experience in T-Sуstems company, where my work is concerned with business analysis and architecture.