It looks like the sizing of the center (both disk space and cpu/memory) are low compared to the actual traffic getting in.
What was the spec recommended by the sizing tool and what was deployed effectively?
was the sizing tool ran for 24 hours covering a full working day?
Any errors in the sizing tool complaining that maybe it failed visibility to some of the DCs (thus missed traffic) ?
In some cases the tool can be off, it's heuristics based on past customer experience, but for some environments you might need to bump the spec up .
If the sizing tool did not report any issues, and hardware was deployed according to recommendation, I suggest to open a support case so the engineer can inspect the data collectors to see what is holding back the center, and if indeed we need to add more hardware in this case.