This is a LOADED question. In this space there is no "Usual" or pattern of implementation. Big data is use case driven and the implementation rules follow the business drivers.
From a purely environment perspective it is recommend to have dev/sandbox, UAT, and prod environment. Your business may require DR and that is time to implement one.
how would you Implement the Development and Test solution? I not sure I understand but I will take a shot here. the dev cycle and how you implement is based on the tech you use..ie hive, pig, spark, mr, storm, etc. There is not one pattern. Each enterprise follows different dev standards. Testing follows same standards. I have worked with enterprise where they leave it to dev teams to build their own test solution. Other enterprise have a extremely rigorous test definition and how test pass (ie code scanner for threat detection).
I hope that helps. Again there is no 1 standard. The implementation is a use case driven and all else follows those rules.