Use Case I: Solution using FASRC

• AWS setup is efficient but expensive to use long term
• FASRC provides a more sustainable long term solution
• Our plan is to install PostGIS and OmniSci as public apps on the FASRC cluster
  • Thereafter, transfer the PostGIS based KNN solution to FASRC
  • Transfer Omnisci based modelling solution to FASRC
• I/O is a big overhead in problems of these scales so loading the data once and utilizing it over and over again makes it more efficient
• Divide and Conquer approach: Currently, multiple AWS instances are launched using the AMI to replicate the database over multiple instances and hence save I/O time
• Replicate AWS: Run several asynchronous jobs on FASRC which will run independent calculations and then combine the results in the end