I suppose you are talking about a VCDX-DCV application.
This could also be fictionary so don't see specific limit in both scenarios.
The main (possible) issue with redesign (or fictionary) projects is that you may not know really deep, or you can make some mistakes (like inconsistence across the documents).
I think that a redesign scenario would be really hard to face... You will have to work with a lot of assumptions from the previous design if you don't have access to the original requirements and design decisions. It may also cause you troubles when you have to justify your new decisions.
Take a look at the VCDX-DCV Blueprint. This is what your design will be marked against and you need to ensure that what you submit and ultimately defend meets all the criteria set out in this document.
I agree that you may not have the insight into the existing network/storage infrastructure to answer low-level questioning about it, which you will of course have to do.
I have to agree with Martin. Look at the blueprint and see which design aligns with it best. I did a brownfield upgrade and redesign with mine and it went great. But it was 100% about aligning with the blueprint not the complexity of the design.
Thanks for all your comments. I'm currently going with the redesign scenario, and aiming for implementing the new design at the same time as the customer is changing to new production hosts in early 2015. At the same time, I'll merge their Prod and Test/dev environments to make better use of the hardware (less 2-node clusters), since lowering TCO is the top requirement.
At the same time, I will clarify the customer's requirements and constraints and redo any old design decisions if I find they are not fulfilling the requirements.
I will study the VCDX-DCV Blueprint closely and make sure I cover everything there.