Test Data Manager

 View Only

CA Service Virtualization meets Grid-Tools – CA Data Finder how will they get along

By Anon Anon posted Jun 06, 2015 09:32 PM

  

I have been getting questions on 'how is test data management different than a virtual service that provides test data management' for many years now. As many of you know, I became a disciple of service virtualization back in 2007 because it gives me so many great things to help with my development, validation and non-functional testing.  Oh service virtualization, you promised me that if observing current behavior it is a prediction of future behavior, you will work just fine.  That you will mind the clock and shift time so that future time is relative to my current preventing my data from becoming old and stale. That route replay of transaction is never enough, those pesky primary keys and correlation of data from request to response will be handled by value.  And finally that you will maintain the state of my data when I ask the same questions of your over and over again; 'what’s my balance, what’s my balance, what’s my balance' and all of those data questions that require state, you will answer with grace and accuracy.


Given all of all this greatness, why do I need anything else?

 

Well unfortunately, a single application is ever so composite in nature. I may be able to virtualize away my troubles with a virtual service so that I can concentrate on my own development, but there is a time my application, functionality, or service has to talk to the others.  This is kind of like the first time you get on the playground and want to make sure you have something in common other than all of us on the playground were dressed by our mothers.  When your application moves to the SIT or is presented at the end of a sprint, who will come play with my application and will our services get along?  I mean a contract is great (thank you RAML and WSDL) but it is all about the data at that point when we start to play.

 

Screen Shot 2015-06-06 at 8.27.49 PM.png

If we walk through the picture, imagine I am writing my simple mobile application and the world is just another JSON document to me. But then I have to learn to play well with others. I move to a sprint boundary or system integration test and all of a sudden there is a real service, not my trusty virtual service.  Then it gets even worse! The closer I get to production the more “real” the systems get and less virtual.  This is a very natural progression, developers want to have stable environments that follow there specification and have all the right data so they can code cool alerts and color in their mobile application.  Then comes in the horrible System of Record for another service and all of a sudden we don’t have the common framework of a virtual service.  These two, the client (mobile application) and the server (customer names, real loan applications), need common data. They need their real systems of records be updated and synchronized.  Having a customer with no loan or a loan without a customer will just stop all progress. Yes virtual service, you solved this for me in the virtual world but now, I need CA Data Finder to understand the relationships of the data between the services and their systems of records. So that they both have common customers and the customers have the right types of and sizes of of loans.  Good thing, CA Data Finder you can build a test data warehouse for me and let me create synthetic data when I can’t extract subset and de-identify from production.

 

Stepping back, I did all of those things in my virtual services but a virtual service is not a replacement for the system of record. Those databases, flat files on Unix and mainframes! I now need data in the ever expanding 'system under test' in a system integration lab or user acceptance area.

 

Oh my clever virtual service, it is true as a developer I want the world to be as virtual and stable as possible. But as a tester in user acceptance testing, I want as much as real as possible with a few exceptions like those 3rd party paid services (like credit checking).  Given I am a tried and true disciple of service virtualization, I do see that having Grid-Tools / CA Data Finder latter in my SDLC is a good thing. The closer I get to production, the more real my systems get and the more I need my systems of record to be more real and coordinated across all of the supporting applications for my simple mobile application.

2 comments
0 views