NetApp Case Studies GumGum Optimizes Ad Targeting with AI and Cloud Infrastructure
Edit This Case Study Record
NetApp Logo

GumGum Optimizes Ad Targeting with AI and Cloud Infrastructure

NetApp
GumGum Optimizes Ad Targeting with AI and Cloud Infrastructure - NetApp Industrial IoT Case Study
Application Infrastructure & Middleware - Event-Driven Application
Infrastructure as a Service (IaaS) - Cloud Computing
Cement
Construction & Infrastructure
Logistics & Transportation
Product Research & Development
Construction Management
Infrastructure Inspection
Cloud Planning, Design & Implementation Services
GumGum, a company that uses artificial intelligence to determine the context of a webpage and match targeted ads, faced a significant challenge in managing its computing infrastructure. The company's proprietary AI services required a large volume of computing infrastructure with variable demand. While AWS cloud was a solution, the costs were high, especially for AI workloads using GPU instances. The company's success led to rapidly increasing costs in line with the demands on their highly scalable infrastructure. The demands were cyclical, with ad serving varying throughout the year. At peak times, GumGum's Ad Server application received over 18 million requests per minute, requiring more than 130 AWS virtual machines to be provisioned. The company's costs grew to around $120k per month for EC2 on just one of their four Ad Server data centers. GumGum needed a solution that would help them use spot instances, save money, be easy for limited staff to use, and guarantee uptime.
Read More
GumGum is a company that uses artificial intelligence to determine the context of a webpage and match targeted ads with the page. The company's AI systems scan images, text, and video on the page to make these matches. GumGum's services require a large volume of computing infrastructure with variable demand. The company's success has led to rapidly increasing costs in line with the demands on their highly scalable infrastructure. GumGum operates four Ad Server data centers, with the largest receiving up to 18 million requests per minute at peak times.
Read More

GumGum

Read More
GumGum turned to Spot by NetApp to meet its needs. Spot offered Elastigroup, a fully managed service for provisioning existing applications on spot instances. The application's ability to predict and preempt spot terminations meant that the availability of applications running on these instances was guaranteed, backed by an SLA. Spot also addressed GumGum's deployment problems. Previously, GumGum was operating on in-place deployment through CodePlay, which resulted in capacity drops and production issues. Using Spot to create a blue-green deployment, GumGum could deploy the entire cluster at once, maximizing capacity and performance, identifying errors more rapidly, and resulting in 50% faster deployments. Additionally, GumGum used Spot Ocean analytics to monitor and right-size its AWS instances. Ocean compared the historical CPU and memory usage of spot instances and predicted future demand based on the past. It also helped GumGum maintain insights into development costs.
Read More
The use of Spot by NetApp has brought significant operational benefits to GumGum. The company has been able to manage its variable demand and high computing infrastructure needs more effectively. The ability to predict and preempt spot terminations has ensured the availability of applications running on these instances. The use of blue-green deployment has maximized capacity and performance, allowing for faster identification of errors. The use of Spot Ocean analytics has enabled GumGum to monitor and right-size its AWS instances, preventing wasted money from oversize instances and poor performance from undersize instances. Spot has also helped GumGum maintain insights into development costs, a typical shortcoming of a containerized development infrastructure.
Saving of approximately $500K annually on infrastructure costs
50% faster deployments with Spot's blue-green deployment
Cost savings from right-sizing AWS instances with Spot Ocean analytics
Download PDF Version
test test