Case Study: PSR achieves massive parallel processing and dramatic runtime reductions with Amazon Web Services

PSR - Customer Case Study

PSR, a Rio de Janeiro–based technology and consulting firm serving utilities worldwide since 1987, builds optimization and simulation tools (notably the SDDP hydrothermal model) that require massive, time‑coupled computations. Their challenge was delivering these models quickly and affordably for customers whose datasets demand far more processing power than a single server or local datacenter could provide, especially during peak runs.

PSR moved its products to AWS, using EC2 Cluster Compute instances and a centralized PSR Cloud client/server architecture (with authorization, queuing, and cloud‑wrapping modules) to spin up, control, and terminate parallel clusters on demand. The cloud enabled dramatic scale and cost efficiency—over 44,000 processor hours in one month (which would have taken 76 days on local infrastructure) and runs using 1,024 processors in parallel—while allowing pay‑per‑use capacity and faster reporting (now exploring Elastic MapReduce).

PSR

Mario Veiga Pereira

President