views:

36

answers:

2

Hi

Can anybody explain the what are the business cases for storage regarding testing environment. Please provide any guidence regarding this.

Thanks,

+1  A: 

Q: "What are the business cases for storage/backup industry"

A: It is often cheaper to backup than loose valuable data.

Aiden Bell
A: 

A proper testing environment must cater for test-retest reliability - that is, once a particular software function has been tested and passed, the test and its results should be retain, firstly as an audit check, but secondly for regression testing.

Testing is a quality check. If you want to certify the quality of any product, one of the ways you can do it is to list the tests carried and the test results. This provides some confidence that the product is "fit for purpose".

Because of this, tests must remain consistent over the years, so that you're assured of measuring the same thing all the time.

This then means that changed functionality must be reflected in the tests, so tests must be updated and versioned in exactly the same way as source code is.

But the major reason for keeping old tests is that source code moves on following a release, and when errors are reported against a version of your software, you need to be able to bring it back and try to reproduce the fault as part of your investigation.

And what applies to test code also applies to test environment. In many cases, a test setup will entail setting up a web server, a database, and service functions needed by the routine under test. The best way to preserve and reproduce this is by using Virtual Machine technology - set it up once, then restore it when you need to test it. This can save a lot of time for large commercial applications, but it is also useful for small scale jobs as well - it relieves the stress (and boredom) of setting a machine up to exactly the same configuration you ran with last time you tested it.

How long should you retain tests? As long as it takes. With large systems (most of my working life) we have had releases out in the field more than two years old, and the source for those tests would have been modified up to twelve months previously., so for us three years would have been a minimum.

The next issue is: How many test sets? Again, this is application-dependent, but it could range into a dozen or a hundred different sets of system definition and test configurations just for one system.

Storage is needed :)

Doug Scott