Tens of thousands of customers use Amazon Redshift for modern data analytics at scale. It is optimized for efficient query processing and automatic tuning. Powerful SQL analytic capabilities on unified data across Amazon Redshift data warehouses and Amazon Simple Storage Service (Amazon S3) data lakes allow customers to enable near real-time analytics to accelerate decision-making. Amazon Redshift Serverless makes scaling analytics effortless, allowing analysis of petabytes of data without the burden of infrastructure management.

In this talk, we focus on a modern multi-warehouse architecture and demonstrate how autonomics and AI are used to optimize cost and performance for our customers. First, we examine how to enable workload isolation, chargeback, and data collaboration working on a single copy of data. Second, we show how machine learning can be used to automatically leverage materialized views. Third, we demonstrate how our recommendation engine analyzes workloads for tuning, and how AI drives automatic scaling and optimization decisions.

This paper introduces EdgeMLOps, a framework leveraging Cumulocity IoT and thin-edge.io for deploying and managing machine learning models on resource-constrained edge devices. We address the challenges of model optimization, deployment, and lifecycle management in edge environments. The framework’s efficacy is demonstrated through a visual quality inspection (VQI) use case where images of assets are processed on edge devices, enabling real-time condition updates within an asset management system. Furthermore, we evaluate the performance benefits of different quantization methods, specifically static and dynamic signed-int8, on a Raspberry Pi 4, demonstrating significant inference time reductions compared to FP32 precision. Our results highlight the potential of EdgeMLOps to enable efficient and scalable AI deployments at the edge for industrial applications.

In modern manufacturing ecosystems, data serves as a keystone for operational efficiency, quality control, and innovation. Additionally, manufacturing industries heavily rely on a clear separation of distinct manufacturing processes. While each single manufacturing process can be arbitrarily complex, the overall production must adhere to certain standards, such that the products exhibits a high level of quality. These standards also extend to the data produced and consumed by the manufacturing processes. In this paper, we focus on changes in modern data management and governance to be more efficient by integrating data contracts into modern data architectures. We show how data contracts are used at ZEISS SMT, a global leader in semiconductor manufacturing equipment. We detail concrete advantages of how data contracts help us find a balance between a centralized and decentralized data management strategy, and thus allows ZEISS SMT to gain momentum when new production processes are established. Furthermore, this paper highlights important lessons learned during the last five years of using data contracts in an industrial use case.