How 605 Reduced EMR Costs by 60% in Less than a Week

Download Case Study

60%

Cost Reduction

30%

CPU Utilization Reduction

60%

Average Job Completion Time Reduction

Granulate performance optimization allowed 605 to run the same EMR jobs 30% faster and with 30% lower CPU utilization. Using Granulate 605 was able to significantly reduce their big-data costs by executing each job faster and with fewer instances. Within less than a week, 605 achieved a 60% cost reduction, 30% reduced CPU utilization, and an average job completion time reduction of 60%.

About 605

605 is an independent TV measurement and analytics firm that offers advertising and content measurement, full-funnel attribution, media planning, optimization, and analytical solutions on top of our deterministic TV viewership dataset covering more than 22 million households across over 200 U.S. markets. 605’s multi-source viewership dataset offers whole-home TV viewing visibility by combining the best attributes of set-top box and ACR data. 605 is unique in that its multi-source viewership dataset supports 100 percent deterministic audience measurement at the household level while being reportable second by second with currency grade projection methodologies, all in a privacy-compliant manner.

Granulate helped us cut our EMR jobs completion times delivering lower costs and better agility without any code changes or R&D efforts

Jared Williams, Director of DevOps

The Challenge

In the past year 605 has seen increased demand for its platform’s offering. Following this spike in traffic, 605 now faces the challenges of scaling up infrastructure; the volume of largely unstructured TV audience viewership data flowing from households across the United States grew fast, significantly increasing their cloud costs. 605 live and time-shifted viewership data from 21M households power the future of TV measurement & attribution, creating massive demand for scale and flexibility from their big-data EMR based workloads. These services run hundreds of different jobs with significant compute requirements thus making the challenge of optimization a complex one to solve.

The Results

Cost Reduction – 60% , Reduced CPU Utilization – 30%, Reduced Average Job Completion Time – 60%
Granulate performance optimization allowed 605 to run the same EMR jobs 30% faster and with 30% lower CPU utilization. Using Granulate 605 was able to significantly reduce their big-data costs by executing each job faster and with fewer instances. In addition to the cost-reduction benefit, these performance improvements allowed 605 to enjoy better agility by accessing and analyzing data faster, supporting company growth.