Edge and Cloud: Finding the Right Balance for Data Science Workloads

Many companies moved to embrace the public cloud with all types of workloads, including data science workloads. Today, many of those organizations are reevaluating if the public cloud is always the best place for these workloads. For some, a closer look often reveals a poor fit for development, debugging, and exploration, with services that overserve their performance requirements at a higher cost than they had anticipated. Other organizations come to realize that early decisions around security have to be rethought because their literacy around these topics has increased.

As a result, some companies have begun the process of reviewing which workloads they should keep in the public cloud, which workloads they should move to private cloud, and which workloads should be run locally on devices such as high-powered, purpose-built workstations. According to a recent IDC survey of IT decision makers (ITDMs), 85% of respondents said their companies are actively repatriating public cloud workloads. The top reasons for doing so are security (19%), performance (14%), and cost (12%).


Share content on email