Warehouse-Scale Computers to Exploit Request-Level and Data-Level Parallelism - Computer Architecture: A Quantitative Approach - page 438

FIGURE 6.3 Average CPU utilization of more than 5000 servers during a 6-month peri-
od at Google . Servers are rarely completely idle or fully utilized, instead operating most of the
time at between 10% and 50% of their maximum utilization. (From Figure 1 in Barroso and
Hölzle [2007] . ) The column the third from the right in Figure 6.4 calculates percentages plus or
minus 5% to come up with the weightings; thus, 1.2% for the 90% row means that 1.2% of
servers were between 85% and 95% utilized.
In summary, WSC hardware and software must cope with variability in load based on user
demand and in performance and dependability due to the vagaries of hardware at this scale.
Example
As a result of measurements like those in Figure 6.3 , the SPECPower benchmark
measures power and performance from 0% load to 100% in 10% increments (see
Chapter 1 ). The overall single metric that summarizes this benchmark is the
sum of all the performance measures (server-side Java operations per second)
divided by the sum of all power measurements in wats. Thus, each level is
equally likely. How would the numbers summary metric change if the levels
were weighted by the utilization frequencies in Figure 6.3 ?
Answer