== Using MapReduce-style data analytics frameworks (5pts) ==
5   Report shows evidence of student's ability to run MapReduce jobs and
        to vary both basic compute (number of slaves) and basic data 
        (replication level) parameters
3   Report shows evidence of student's ability to run MapReduce jobs and
        to vary either basic compute or basic data parameters 
1   Report shows evidence of student's ability to run MapReduce jobs with a 
        default configuration
0   No MapReduce jobs were run
== Effects of cluster size (2pts) ==
2   The effect of cluster size on MapReduce performance is accurately
        explained and supported by measurement results. 
1   Measurement results are presented for varying cluster sizes, but the
       results are not explained accurately or at all. 
0   No results or explanation of cluster size effects is included
== Effects of data parameters (4pts) ==
4  All of the following 3 questions are answered with insightful, accurate
        explanations supported by measurement results:
        How does the performance change with different block sizes?
        Does the replication level influence performance?
        How do different types of MapReduce jobs respond to the same
            replication level?
3   2 of the above questions are answered with insightful, accurate
        explanations supported by measurement results OR results are presented
        for 3 of the above questions but the results are not sufficiently
        explained
2   1 of the above questions is answered with insightful, accurate
        explanations supported by measurement results OR results are presented
        for 2 of the above questions but the results are not sufficiently 
        explained
1   Results are presented for 1 of the above questions but the results are 
        not sufficiently explained
0   No results or explanation of data parameter effects is included
== Effects of compute parameters (4pts) ==
4  All of the following 3 questions are answered with insightful, accurate
        explanations supported by measurement results:
        How does the number of map tasks influence performance?
        How does the number of tasks per slave influence performance?
        How does the number of reduce tasks influence performance?
3   2 of the above questions are answered with insightful, accurate
        explanations supported by measurement results OR results are presented
        for 3 of the above questions but the results are not sufficiently
        explained
2   1 of the above questions is answered with insightful, accurate
        explanations supported by measurement results OR results are presented
        for 2 of the above questions but the results are not sufficiently 
        explained
1   Results are presented for 1 of the above questions but the results are 
        not sufficiently explained
0   No results or explanation of compute parameter effects is included
== Performance estimation (3pts) ==
3   Differences between medium and high-cpu medium instances that may
        influence MapReduce performance are discussed; Values are specified
        for all compute and data parameters; Value choices are thoroughly
        explained and backed by earlier observations 
2   Differences between medium and high-cpu medium instances that may
        influence MapReduce performance are discussed; Values are specified
        for all compute and data parameters
    OR Values are specified for all compute and data parameters; Value 
        choices are thoroughly explained and backed by earlier observations 
1   Values are specified for all compute and data parameters
0   No MapReduce jobs are run on a cluster of high-cpu medium instances or
        parameter choices for this setup are not specified
== Parameter critique (2pts) ==
2   Choice of parameters is critiqued based on observed performance on a
        cluster of high-cpu medium instances; New values are specified
        for at least one compute or data parameter
1   New values are specified for at least one compute or data parameter
0   No second round of MapReduce jobs are run on a cluster of high-cpu medium
        instances or parameter choices for this second round are not specified
== Choice of performance metrics (3pts) ==
3   Several metrics of performance are considered; An appropriate
        subset of metrics are considered when evaluating specific parameters
2   Several metrics of performance are considered; Metrics considered do 
        not vary based on the parameters being evaluated
1   Only one metric of performance is considered for each parameter being 
        evaluated
0   Only one metric of performance is considered throughout the entire study
== Measurement results (2pt) ==
2   All measurement results are presented in a clear format and accurately 
        labeled.
1   Some measurement results are not presented in a clear format or accurately
        labeled.
0   Measurement results are hard to interpret