Assume that the MapReduce job from Page #438 in Section 6.2 is executing a task with 2^40 bytes of input data, 2^37 bytes of intermediate data, and 2^30 bytes of output data. This job is entirely memory/storage bound, so its performance can be quantified by the DRAM/Disk bandwidth of Figure 6.6.

a. How much does the job cost to run on m4.16xlarge and m4.large in Figure 6.15?
Which EC2 instance provides better performance? Which EC2 instance provides better cost?

b. How much would the job cost if an SSD was added to the system, as in m3.
medium? How do the performance and cost of m3.medium compare with the best instance from part (a) above?

Figure 6.15:

General-purpose Compute-optimized Instance 12 nano 12.micro 12 small 12.medium 12.large 12.xlarge 12.2xlarge m4.large m4.xlarge m4.2xlarge m4.4xlarge m4.10xlarge m4.16xlarge m3.medium m3.large m3.xlarge m3.2xlarge c4.large c4.xlarge c4.2xlarge c4.4xlarge c4.8xlarge e3.large e3.xlarge c3.2xlarge c3.4xlarge c3.8xlarge Per hour $0.006 $0.012 $0.023 $0.047 $0.094 $0.188 $0.376 $0.108 $0.215 $0.431 $0.862 $2.155 $3.447 $0.067 $0.133 $0.266 $0.532 $0.100 $0.199 $0.398 $0.796 $1.591 $0.105 $0.210 $0.420 $0.840 $1.680 Ratio to m4.large 0.05 0.11 0.21 0.4 0.9 1.7 3.5 1.0 2.0 4.0 8.0 20.0 31.9 0.6 1.2 2.5 4.9 0.9 1.8 3.7 7.4 14.7 1.0 1.9 3.9 7.8 15.6 Virtual cores 1 1 1 2 2 4 8 2 4 8 16 40 64 1 2 4 8 4 8 16 36 2 4 8 16 32 Compute units Variable Variable Variable Variable Variable Variable Variable 6.5 13 26 54 125 188 3 6.5 13 26 8 16 31 8 132 7 14 28 55 108 Memory (GiB) 0.5 1.0 2.0 4.0 8.0 16.0 32.0 8.0 16.0 32.0 64.0 160.0 256.0 3.8 7.5 15.0 30,0 3.8 7.5 15.0 30.0 60.0 3.8 7.5 15.0 30.0 60.0 Storage (GB) EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only EBS only 1x4 SSD 1 x 32 SSD 2x 40 SSD 2 x 80 SSD EBS only EBS only EBS only EBS only EBS only 2 x 16 SSD 2 x 40 SSD 2x 80 SSD 2x 160 SSD 2x320 SSD