Modernising Legacy Applications for SOA Environments

By Nick Hortovanyi

How do we start to realise the potential of an SOA style architecture? A great start would be modernisation of existing legacy systems, to expose functionality that can be potentially consumed in composite applications or mashups.

Because of the large capital investment in legacy systems, it is preferable to find a satisfactory migration path from the legacy system environment to modern SOA (Service Oriented Architecture) environments. 

Somewhat contentious, would be to decommission existing systems in favour of a SaaS service or to use a packaged COTS application. The following discussion is for when this is not an option.

Legacy applications need to be modernised in a fashion that automates the majority of "eyeball" type work. Current best practices suggest a Model Driven Development approach with appropriate modern tools.

I've identified three high level options for migration of legacy applications into a SOA environment, with some identified advantages/disadvantages.

Option 1 – do-nothing

-          Staff retention – people no longer wish to work on legacy applications – bad career choice for future

-          Attraction of new talented staff – old generation technology, staff want to work on Projects utilising modern development tools and methodology

-          High cost to change to adapt to new business processes

-          Expensive runtime licenses in addition to licensing for new environments

-          Unable to meet business expectations to respond in a timely manner

-          Difficult to integrate with other systems, very rigid in purpose

-          Large sites, have many generation of technolog needing specialist knowledge of each generation

-          Unable to change platform of choice for middleware and databases

Option 2 – manual

-          Labour intensive due to the need to eyeball code and document specifications for new systems

-          Open to error, security problems and inconsistent coding

-         Offshore code factories, whilst reducing some short term costs do not solve architectural issues and still retain the other problems associated with manual transformations

-          Often insufficient budget to achieve the targeted strategic architecture for the business

-          Unable to reengineer, without a complete rewrite – often the original analysts and designers are no longer available to assist

-          Timeframe are usually long >1 year for projects, resulting in a freeze of all change and hence responsiveness to business

Option 3 – tools

-          Reduces 85% of the eyeball work required

-          Reduction in timeframe target. In general, it is expected to achieve, a quarter of the timeframe for manual methods

-          Through using a model driven approach ability to refactor components for modern architectures

-        Ability to introduce new elements such as Web Services to comply with the strategic architecture

-          Reduce overall project costs and the number of staff required to complete

-          As the majority of code is generated, it ensures code conforms to Best Practices and Design patterns. Thus reduces coding errors.

-          Improves security, through consistent code generation of high quality code (no shortcuts taken by developers to meet timeframes)

-          If appropriate tools are selected, they can also be used for new application development

-          Improves dramatically the ability of IT to help facilitate changes in business processes through the use modern paradigms such as SOA.

Tags   |  Comments 2