Section 6.7 discussed the use of per-server battery sources in the Google design. Let us examine the consequences of this design.

a. Assume that the use of a battery as a mini-server-level UPS is 99.99% efficient and eliminates the need for a facility-wide UPS that is only 92% efficient.
Assume that substation switching is 99.7% efficient and that the efficiency for the PDU, step-down stages, and other electrical breakers are 98%, 98%, and 99%, respectively. Calculate the overall power infrastructure efficiency improvements from using a per-server battery backup.

b. Assume that the UPS is 10% of the cost of the IT equipment. Using the rest of the assumptions from the cost model in the case study, what is the break-even point for the costs of the battery (as a fraction of the cost of a single server) at which the total cost of ownership for a battery-based solution is better than that for a facility-wide UPS?

c. What are the other trade-offs between these two approaches? In particular, how do you think the manageability and failure model will change across these two different designs?