Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
Phase Three - Model Maintenance
After all of the hard work involved in building a model and ensuring that it works correctly and accurately predicts the system, you will often find that the system will be:
- Upgraded with new hardware or software that needs to be characterized
- Additional and/or modified workload requirements will be added to the system
Both of these situations are quite common, and if you were successful in building the model in the first place, management will hopefully ask you to revise the model rather than making one of the typical requests:
- Use your engineering judgment and just go buy the new hardware
- Just install the hardware/software -- the vendor says it will work
- Throw a dart and pick something
The steps to model revision are the same as developing the model, except that you have many of the components already characterized and you already have a track record for success.
In my 22 years in the computer industry I have seen modeling used infrequently, but always to great success. Scientific modeling is part of our everyday life and is used in almost everything we eat, drive, and fly, and every medicine and vaccine we take, with the exception of the one thing that makes it all possible -- the computer systems that these scientific models run on.