Load Google Translate How do you stop the cost of test escalating?
Testing is about risk mitigation to the business and increasing customer satisfaction/ stock market confidence but how do you stop the cost of test escalating?
A 'Rule of 10' in Electronic Test which means that at every stage, the cost usually goes up 10 times.
If test does not sit in the early stages & major issues found later then the financial effect can be catastrophic.
How have you avoided this?
Actually Tom, you answered your own question. It is all about risk management. If you reuse trusted parts and designs, the amount of test required goes down. If you use new designs and unknown components the tests increase. It is a very simple equation, but even with that said, you need to design and build a testing station for your part of the system. If you are the system integrator, then you build the system level test and demand that each component be tested prior to integration.
So that gets us back to my first point. Reuse! It is the only way to reduce your development testing costs reguardless if it is harware or software. If you trust the component works, through prior testing, then you have low risk in using it. You need to have the risk and testing cost discussion at proposal time. Otherwise, you will always underestimate the effort needed to test and verify your design.
Most successful companies stress reuse at all times. Yes, you may not get the most capable system, but if you can get a system that meets specifications and works, then you are much further ahead than providing a system that mostly works, some of the time.
Even in the proposal stage, I would always try to ask the question about "How are we going to test the requirements?" That is how early you have to start.
I have worked for companies that only bid the development costs and then expected us poor engineers to find some way to cheaply test the product. That approach almost always failed.
Even with reused components, you need to do a level of validating them to ensure that they indeed do what you need them to do. Some people get so convinced that they have no risk due to reuse that they cut out some critical testing. The best example of this approach was the Hubble Space Telescope. They waited until it was in orbit to check the optics. Instead of a simple $10 million test on the ground, they ended up with $500 million Space Shuttle mission to put in a simple set of correction lenses and mirrors.
In systems engineering your motto must always be "Trust but verify!"
Thanks,
DAB
© 2009 Premier Farnell plc. All Rights Reserved
Premier Farnell plc, registered in England and Wales (no 00876412), registered office: Farnell House, Forge Lane, Leeds LS12 2NE