Identify the requirements that are the absolute minimum to allow the product, feature, or subsystem to function for the major stakeholder needs. Focus on delivering only those requirements in early iterations, and non-critical features in subsequent iterations.
Ensure developers, architects, stakeholders and managers are dedicated to delivery of the product. Enable them to do their job or task without interruption, and trust their judgement to do what’s right.
Frequent smaller delivery helps to keep the stakeholder engaged and the development team informed of changes in the business climate. The efficiency of working in small batches is one reason why adopting agile development has benefited the software industry.
A handoff occurs when one team member hands off a piece of work to another (for example, a developer fixes a defect and hands it off to the tester to verify the fix). Flow is the process by which a team member can complete a task, and that the system can move through the development cycle from ideation to production.
The idea that working in small batches is a far more efficient than working in large batches is nothing new. The efficiency of working in small batches is one reason why adopting agile development has benefited the software industry. In the days of waterfall development, months of coding were followed by months of testing, and only near the end of the testing phase did anything really work.
Good communication improves the collaboration level between teams. High levels of transparency significantly reduces the costs and waste through the app delivery. The risk of using many different tools will be the problems integrating them. Solving those problems create noise and need to be one of the most important features of a DevOps toolchain orchestration solution.
Find process steps that add little to no value to the business and remove them. Look at complex approval cycles, and manual steps. Enable team members to make non-critical decisions. Identify communications barriers, and zealously remove them.
Automated testing can quickly identify regression errors quickly. Tests should be automated at the unit, API, and user interface layers, and run before every deployment. Manual deployment steps are error prone, and time consuming. Automated deployments are repeatable, predictable, and auditable (and reduce the team’s anxiety at deployment time).
Still using 3270 ISPF sessions to edit source members? Modern integrated development environments (IDE’s) are desktop based, have rich editors, can scroll endlessly, and have advanced debugging, code coverage and unit testing abilities.
Legacy SCM’s cannot compare to modern systems such as Git, or Team Concert. Rich streaming, and private branching make atomic changes much easier to make, debug, and test.
Need to schedule deployment of a CICS/DB2 application at the same time as a WebSphere app? No problem. Modern tools can do this, and can schedule deployment for off-hours convenience using fully automated, traceable, auditable processes.
Tired of having to retest an entire system because of one line of code change? Automation can make that a breeze. Whether it is using zUnit testing, automated functional testing, or virtualizing subsystems such as CICS or DB2 to isolate testing to only affected load modules, automation of tests can vastly improve your time to market as well as reduce time for regression testing.
