Infusing artificial intelligence (AI)and machine learning (ML) into data and applications at the edge has the potential to transform operations at Federal civilian agencies and in national defense. But edge environments can be challenging to manage and come with their own set of security risks.Meri Talk recently sat down with Cornelia Davis, technology fellow and vice president of product at Spectro Cloud; Tommy Scherer, principal architect at Spectro Cloud Government; and Pragyansmita Nayak, chief data scientist at Hitachi Vantara Federal, to discuss accelerating edge innovation and securely managing edge computing environments.
I see four key challenges: bespoke environments, scale, sporadic connectivity, and security. Bespoke environments require a lot of manual labor and customization. The solution there is Kubernetes, which has become the operating system in any environment. It helps mitigate the software-related challenges of managing and scaling edge applications and application delivery. As for scale, the government might be managing hundreds, thousands, or tens of thousands of environments. The answer there is automation and centralized management. And sporadic connectivity – military applications are often in areas with no internet connectivity, but they still need to work. Application management is centralized, but control must be distributed. And lastly, devices are in the field, perhaps in hostile territory. They need to be secure.
The key challenge, as Cornelia rightfully pointed out, is the distributed nature of working and the decentralized infrastructure. Data is generated in different formats, on different types of devices, and even on different versions of the device. You’re getting your computation and your data closer to where it is needed, but data lineage and data provenance become increasingly difficult. Connectivity is another data-centric challenge. In the future, if you want to tap into the data created at the edge, it needs to come back to the cloud. Lack of reliable communications is a hindrance.