I am excited to announce the general availability of Azure Batch, our job scheduling and compute pool management service.
Azure Batch helps developers easily scale their compute-intensive workloads to tens, hundreds, or thousands of virtual machines without having to manage the infrastructure. In a world of rapidly evolving products and fierce competition, our goal is to deliver a service that lets you focus more on your applications and less on plumbing. As a managed service, Azure Batch handles the heavy lifting of provisioning, monitoring and scaling virtual machines. You create an Azure Batch account, within minutes have the resources you require, and can scale up and down as the volume of jobs and tasks change. Batch helps you handle spikes; you pay for what you use.
It’s rewarding for us to see the variety of workloads and industries where Azure Batch is already being used. From manufacturing, life sciences, financial services, and entertainment, Batch helps organizations work at scale and deliver solutions for their users and customers.
With a rich API for managing compute pools, the processing of jobs and tasks can happen on demand, on a schedule, or as part of a business workflow using Azure Data Factory. Azure Batch was designed from the ground up to handle hyper-scale. We take advantage of this scalability internally to test Azure itself.
GA Updates
With the general availability of Batch, we are also announcing a price change to the service. The resource management and job scheduling capabilities will be free. You will not pay an additional cost for running jobs above the charges for compute, storage and networking resources that your application uses.
Based on customer feedback from the preview, we are introducing a new API, unifying the Batch and Batch Apps namespaces released at preview. The core job and task API is being introduced now. The Batch documentation, client library and samples will be updated and expanded over the coming days. We’ll provide guidance on how to use and migrate to the new namespace.
The management experience for Batch is moving to the new Azure portal. You can monitor pools and jobs, and we will be delivering regular updates with additional capabilities. The Batch Apps capabilities such as a job splitter and task processor remain in preview, and will be included in the new API through regular updates over the next few months.
Customers Using Batch Today
Since the preview release of Azure Batch last year, we’ve been working closely with early adopters to make sure we have a scalable and reliable service. These customers are using Batch today for their production workloads.
Towers Watson, the world’s largest provider of actuarial software, chose Azure Batch to quickly bring to market a new cloud service for their RiskAgility FM application that enables life insurers to run financial models that accurately reflect their products, risk, and capital exposure. This service opens up new opportunities for Towers Watson and their customers.
“We have a long history in HPC, moving to Azure Batch was a breath of fresh air. No infrastructure on premises, everything is in the cloud, it automatically scales up and scales down with an auto-scale formula that is very flexible, and you can define how you want your nodes to start and stop – you can have them be always available or spin them up on demand.” – Cameron Murray, Senior Consultant, Towers Watson
Azure Batch is also ideal for media workloads such as rendering and transcoding. Access to compute on demand helps service providers efficiently scale to meet the needs of their customers. Azure Batch works behind the scenes for Microsoft services, providing tens of millions of core hours per month for teams like Azure Media Services and Xbox Video.
TVEverywhere is a Microsoft partner in the UK that helps organizations make the most of TV and video in the cloud. Their new service VidCoding, built on Azure Batch, is an automated cloud based encoding platform that offers high throughput and low cost to prepare content for distribution on the web, mobile or to the broadcast market.
“Using Azure Batch we are able to scale up and down from very little activity to handling large volumes of encodes from clients across the world in a very cost efficient manner, we can also optimise our service by selecting the appropriate machine instance for every encode depending on the size of the task and the priority of the job.” – Iolo Jones, CEO, TV Everywhere
Our team is excited about this release, and is working hard to add new capabilities such as support for MPI applications and Linux virtual machines later this year. We look forward to hearing what interesting applications you have built on top of Azure Batch.
As always, please send us any feedback, comments and any questions you may have through the comments below, the Batch forum or by contacting us directly here.
Thank you!