Major global electric utility provider delivers modern smart meter data lakehouse powered by AIStor

Our customer, a large electric utility company with millions of users, was faced with challenges common to utility and energy companies around the world: how to manage ever-growing energy demands, while diversifying energy supply to incorporate renewable and sustainable sources. A key element of their strategy is a smart metering initiative - bringing the power of technology and AI to intelligently adjust energy consumption in each home. This promised to help smooth the spikes in energy demand, reduce the likelihood of brownouts, reduce costs to consumers, and improve operational efficiency and productivity for our customer.

Environment: A modern solution for evolving energy demands  

The electric utility company selected the Impresa.ai Meter Data Management system as the overall platform to drive their smart meter initiative. A key piece of the project was to identify a suitable data lakehouse able to support Impresa.ai, providing performance, scalability and throughput capabilities to enable initial implementation and future growth.   

Challenge: Legacy technology challenges and delays

Since the company had been utilizing Hadoop for several years, their initial plan was to expand this environment in order to provide the S3-compatible storage layer required by Impresa.ai. However, it was determined that 240 additional servers would be required to support both the initial launch requirements and anticipated growth, and that implementation would take upwards of a year.  Operational support of this platform would also require 3 full-time FTEs. This was simply incompatible with the customer’s project goals, budget and timelines.   

Solution: AIStor-powered data lakehouse 

Choosing AIStor as the data lakehouse underpinning Impresa.ai meant that this utility was able to meet their critical business deadlines significantly faster, realize their business objectives more rapidly, while also positioning themselves for additional future analytics and AI use cases.

AIStor is software-defined, enabling compute and storage to be decoupled. This enables each tier to be scaled based on the needs of the use case, driving significantly improved infrastructure efficiency. This is also a key factor enabling the rapid implementation of AIStor, as a simple-to-deploy binary on commodity hardware.  

AIStor is designed from the ground up to focus on speed. Factors such as being a single layer solution, rather than a multi-layer or multi-node stack, having no metadata database as a single point of failure or bottleneck, and leveraging Assembly Language and SIMD acceleration, all contribute to AIStor’s ability to deliver query response time far that far exceeds the legacy HDFS environment.

Outcomes: In production faster with lower TCO

A hardware footprint of 90 servers compared to 240 meant reduced cost, power and complexity, while delivering better performance than legacy Hadoop-based solutions. This resulted in a more than 50% lower Total Cost of Ownership (TCO). 

AIStor’s architectural simplicity meant that production implementation and deployment was completed in just 10 weeks, compared to the 14 months estimated for a Hadoop-based solution - an 86% Faster Time to Production.

AIStor’s operational simplicity also means that it requires less than 1 FTE to support and manage, a significant reduction compared to the staffing overhead associated with the legacy Hadoop stack, delivering more than 66% time and resource savings.