ADVSUS01-BP01 Distribute data and workloads across Regions when necessary to minimize network usage and latency
When selecting regions to host workloads for sustainability, distribute data and workloads across multiple Regions to minimize network usage and latency, prioritising the most sustainable Regions available that leverage renewable energy sources. The millisecond latency of programmatic advertising workloads typically requires ad-servicing architectures be near consuming workloads. However, there is opportunity to consolidate data analysis for these workloads into fewer Regions.
Implementation guidance
-
Identify the latency requirements for your workloads, and determine which AWS Regions can meet those requirements.
-
From the eligible regions, select the one with the lowest carbon footprint, considering factors such as the energy mix (prioritize Regions with 100% renewable energy).
-
Use AWS tools to measure and report your carbon footprint.
-
Consolidate infrastructure needs for analytics workloads (real-time bidding, privacy-enhanced data collaboration, ad intelligence, and measurement) in fewer AWS Regions with 100% renewable energy.
-
Use AWS services designed for energy efficiency, such as Amazon EBS gp3 volumes, Amazon EC2 Instances with AWS Graviton processors, and Amazon EC2 Tranium and Inferentia instances for AI workloads.
-
Periodically review and optimize the regional distribution of workloads as new, more sustainable AWS regions become available, balancing sustainability goals with performance requirements.
-
Aggregate analytical data in local regions and move the aggregates to the central reporting region when data needs to be centralized for business reasons.
Key AWS services
