Figure 6

Scaling FIGG with MapReduce. The mutation process in FIGG is the most computationally intensive job in the pipeline. It was tested on Amazon Web Services Elastic MapReduce clusters of varying sizes for scalability. MapReduce provides a near linear speed up with the addition of nodes to this job. These genomes are saved to HBase to provide a persistent store of standard artificial genome data that can scale along with the cluster size. This is one area where optimization will provide increased performance as defining how the HBase tables are distributed can increase the speed of computation (e.g. more efficient row key design decreases query time and increases the number of available mappers). This is due to the fact that region server optimization is highly specific to the data, and improves as the data size increases.