|Google Anthos Adds Hybrid AI|
|Written by Kay Ewbank|
|Thursday, 10 September 2020|
Google has announced improvements for Anthos, its application packaging and management platform for hybrid cloud. The updates include hybrid AI capabilities, support for more cloud providers, and a bare metal beta for running Anthos locally without a hypervisor.
Google Anthos is a platform that can be used to run applications without modification locally or in the public cloud. The idea is that you write an application once, package it into a (Kubernetes) container and then deploy it wherever - Google Cloud Platform, Amazon Web Services, Azure or locally.
Although Kubernetes is now open source, it was originally developed by Google. The same development team created Google Kubernetes Engine (GKE) a cloud-based Kubernetes service for running containerized applications.
In practical terms, Anthos is the GKE deployed locally with an encrypted connection to Google Cloud Platform (GCP) managed by a locally installed agent. The agent can be used to manage your Anthos cluster and the apps you're running on it, no matter what cloud or clouds the cluster is running on. Anthos is used for Google's BigQuery Omni analytics product that can be used to access and analyze data stored in Google Cloud, Amazon Web Services and Azure.
The main improvement to Anthos is the addition of hybrid AI capabilities for Anthos. What Google means by this is that organizations will be able to run apps that include AI elements within Anthos "wherever your workloads reside". The thinking is that AI models usually rely on large amounts of data, which may well be sensitive and has to be located within the organization's data center rather than the cloud. This limits the ability to run AI-based analysis from the cloud because the data can't be uploaded.
The new "hybrid AI capability" in Anthos means you can use cloud-based applications but run them locally to comply with data privacy regulations.
The first AI application from Google in Anthos is Speech-to-Text On-Prem. This gives an organization full control over speech data that is protected by data residency and compliance requirements, from within the company's own data center. The app uses speech recognition models developed by Google's research teams that are more accurate, smaller, and require less computing resources to run than existing solutions.
Google has also added an Identity Service to Anthos, which works with existing identity solutions on Anthos workloads. The initial version of the service supports OpenID Connect on premises and in beta for Anthos on AWS). Support for other protocols will be added in the coming months.
The newly announced improvements also include support for Anthos attached clusters, which let you manage any Kubernetes cluster with the Anthos control plane, along with the beta of Anthos for bare metal, which lets Anthos run on-prem and at edge locations without a hypervisor.
or email your comment to: firstname.lastname@example.org
|Last Updated ( Thursday, 10 September 2020 )|