Microsoft is making its Azure cloud the place enterprises can run their AI and machine learning workloads.
“Data scientists in organizations adopting Snowflake as their data warehouse solution can now explore the capabilities of Azure ML without relying on third-party libraries or working with data engineering teams. said Amar Badal, senior manager of Azure Machine Learning, in a blog post.
He added that thanks to Snowflake’s native integration with Azure Machine Learning, data scientists can “import data from Snowflake to Azure ML and start machine learning projects with a single command.”
Redmond argued that the connection between the two platforms would allow data scientists to save time moving data to Azure on a schedule or on demand, and track where that data moved.
The announcement comes a week after Microsoft enhanced Azure Machine Learning at its Build 2023 conference with an expanded partnership with Nvidia and released a public preview of the underlying model for cloud services.
It’s clear that Microsoft is making moves to beef up its AI training capabilities, but it remains to be seen how well that will work.
The software giant’s latest move is to make it easier for organizations to bring information from data repositories that aren’t part of the Azure platform (think database company Snowflake or Amazon Web Services’ S3 service) into Azure Machine Learning for AI training. includes the introduction of tools that allow importing into At the same time, Redmond is enhancing his suite of tools used to track and manage his AI training jobs with Azure Machine Learning.
The cloud offers flexible access to hardware tailored to AI workloads, but often not the kind of kit that is affordable or comfortable to run on-premises, leaving enterprises We see hyperscalers as a cost-effective way to embrace the rapidly accelerating AI trend driven by Large Language Models (LLM) and generative AI applications like ChatGPT.
Therefore, Microsoft wants to be the go-to cloud for AI workloads, along with AWS, Google Cloud, and other providers. Armed with generative AI products from its multi-billion dollar investment in OpenAI, Redmond is on a multi-month sprint to push machine learning into every part of its software portfolio.
This new initiative is aimed at users of Snowflake’s cloud-based data warehouse.
In this context, Microsoft is exposing lifecycle management capabilities in public preview as a way to manage datasets imported into Azure Machine Learning datastores, or so-called HOBOs (hosted on behalf of). . This is for data imported via CLI and SDK.
“When you select the HOBO datastore as your data import destination, you get access to lifecycle management, a feature called ‘auto-delete settings’ for imported data assets,” Badal writes. “A policy is set on all imported data assets in the AzureML managed datastore to automatically delete the imported data asset if it has not been used by a job for 30 days.”
A new tracking tool is also in public preview and aims to help businesses manage their training tasks. It includes customizable dashboard views with charts that visualize resource usage, ratings, and job metrics to get a more detailed overview of your projects.
There is also a customizable list of training jobs that displays the names of the tasks you are working on. Data scientists can also select and sort columns, filter jobs by various criteria, and perform batch actions on jobs.
Other new features include the ability to compare training project metrics and images, add markdown to notes, and create and save custom views.
Microsoft is clearly committed to AI. ®
