Run ultra-low latency advertising workloads with cloud-only optimizations and unmatched cost-performance
How four leading businesses optimized real-time advertising performance
Achieving optimal price performance and accelerating innovation with real-time advertising on AWS
With unmatched opportunities to optimize compute performance and costs, AWS is the cloud standard for advertising platforms that buy, serve, and measure hundreds of billions of ads per day at single-digit millisecond latency. Customers can move fast and spend less with the broadest set of solutions to migrate real-time advertising workloads, expand into new markets, and improve development velocity with cloud-native best practices and architectures. Customers can achieve up to 50% reductions in unit costs per billion requests in real-time advertising and up to 70% reductions on advertising analytics workloads.
Leading advertising platforms on AWS
The Trade Desk shifted their real-time bidding strategy to the cloud in 2019, building 3 real-time bidding sites on AWS. The company cut production deployment to 1-3 weeks, compared to 3-6 months for a physical data center. The company processes 30 million events per second on AWS for real-time advertising with m5d.metal and i3en instances running Aerospike for its cold storage of user profiles.
"The elasticity that we get from using Amazon EC2 has been really helpful for allowing us to scale up and down as we need to.”
Matt Cochran, Director of Engineering - The Trade Desk
Integral Ad Science processes 100 billion+ ad verification events a day & trillions of events per month, providing real-time recommendations for bidding and context, responding to API calls in 10–50 milliseconds. Until 2019, IAS operated data centers globally, buying its own hardware requiring months to provision.
IAS migrated all-in on AWS in 2020, moving its global edge services first, followed by reporting platforms; ETL pipelines; CI/CD; event collection; and data science tooling. IAS reduced costs by 12% and reduced time to provision new instances from 6 weeks to a few minutes.
NextRoll processes more than 150 billion requests per day and ingests 600 terabytes of data daily with AWS for real-time advertising. NextRoll migrated Amazon EC2 instances with AWS Graviton processors.
The company saw “up to 50 percent total cost savings over our previous generation EC2 instances.”
Valentino Volonghi, CTO - NextRoll
Nielsen Marketing Cloud processes 250 billion ad measurement events per day using serverless computing with AWS Lambda and containers with Amazon EKS to achieve 20 percent cost efficiency.
“AWS helps us democratize the work we do in the measurement space and provide access to all advertisers and platforms, no matter how big or small they are.”
Matthew Krepsik, Global Head of Analytics - Nielsen
GumGum switched to Amazon EC2 Spot Instances and Amazon EC2 G4 Instances with up to 100 Gbps of networking throughput for its machine learning workloads. The switch enabled 4x faster inferences and made for a smaller model, enabling more inferences in parallel.
“Anything we launch these days, we design to run on Amazon EC2 Spot Instances. By moving compute to Amazon EC2 Spot Instances, GumGum has saved 62 percent on compute costs.”
Vaibhav Puranik, Senior Vice President of Engineering - GumGum
AppsFlyer manages, measures, identifies, and blocks fraudulent installs with its Protect360 antifraud solution powered by AWS, using Amazon EC2 Spot instances to run machine learning to process 100 billion events per day, saves its customers $8.1 million per day by preventing fraudulent activity.
“If we reach a certain threshold, we can provision more Amazon EC2 Spot Instances immediately, and we’re able to handle the load."
Ido Berkovitch, Research and Development Director - AppsFlyer
Optimize cost performance
With unmatched compute capabilities, with more sizes and types of instances, along with the fastest processors in the cloud with AWS Graviton2 instances, AWS enables more opportunities to optimize costs for high-performance, high-scale, ultra-low latency advertising platforms. Industry customers can reduce unit cost per billion requests in real-time advertising by up to 60% using cloud-native architecture and AWS compute capabilities such as EC2 Spot instances, which enable up to 90% savings on On-Demand instances.
Industry leaders such as The Trade Desk use AWS to accelerate development of high-performance, ultra-low latency workloads such as real-time bidding by expanding into new markets in days instead of months. AWS also offers purpose-built solutions for real-time advertising workloads that enable cost-optimization and reduce heavy lifting through cloud-native architectures and services such as Graviton2 instances, Amazon EC2 Spot instances, Auto-Scaling groups, and serverless capabilities.
With AWS, advertising platforms accelerate revenue growth through the broadest set of cloud partner opportunities—including inventory and data integrations with Amazon Advertising, as well as distribution of audience and measurement data through AWS Data Exchange, which gives data providers a secure and efficient distribution channel to reach millions of AWS customers. With the largest community of partners and tens of thousands of data lakes running on AWS today, advertising platforms also enjoy greater flexibility and agility on AWS when sharing, enriching, and activating audiences with media partners on the most widely adopted cloud.
Solutions & services
Accelerate deployments and reduce heavy lifting with AWS Partners for Ad Platforms.
Aerospike is a real-time data platform optimized for predictable performance at petabyte scale.
Databricks Lakehouse Platform Databricks on AWS unifies all your data sources into a simple, open, and collaborative foundation of your Lakehouse.
Granulate offers a real-time continuous performance optimization solution to drastically improve performance for any app with no code changes required.
Snowflake has developed a completely new SQL data warehouse, designed from the ground up for the cloud.
Spot by NetApp automates and continuously optimizes cloud compute infrastructure, helping companies reduce their compute costs by up to 90%.
Upsolver gives organizations a simple way to rapidly develop and maintain a world-class data lake, saving hundreds of development hours and thousands of dollars.