Spreading the load: How Salesforce met Multi-AZ HA with SageMaker Inference Components
Salesforce solved multi-region AI inference without the cost tax—here's how they did it.
Why it matters
A practical case study in using SageMaker's scheduling features to balance high-availability compliance with model co-hosting efficiency. Relevant to practitioners deploying inference at scale.
The key facts
8 to knowSalesforce deployed across multiple Availability Zones using SageMaker Inference Components
Used SchedulingConfig parameter for model placement strategy
Maintained cost efficiency through multi-model co-hosting while meeting Multi-AZ HA requirements
This is a vendor (AWS) technical deep-dive on production inference patterns
Salesforce deployed models across multiple Availability Zones using SageMaker Inference Components
Used SchedulingConfig parameter for AZ-aware placement
Maintained multi-model co-hosting cost efficiency while meeting Multi-AZ HA compliance
Published by AWS on SageMaker ML blog
Go to the source
AWS Machine Learning Blogaws.amazon.com
Publisher excerpt: Learn how Salesforce used Amazon SageMaker AI Inference Component placement (the SchedulingConfig parameter) to distribute model copies across multiple Availability Zones, meeting their Multi-AZ high availability compliance requirements without sacrificing the cost efficiency of multi-model…