Microsoft Azure already solidified its position as the second most popular public cloud, and critical additions in 2017 brought the Azure feature set closer to parity with AWS.
In some cases, Azure leapfrogged its competition. But a bevy of similar products bolstered the platform as a viable alternative to Amazon Web Services (AWS). Some Microsoft initiatives broadened the company’s database portfolio. Others lowered the barrier to entry for Azure, and pushed further into IoT and AI. And the long-awaited, on-premises machine, Azure Stack, seeks to tap surging interest to make private data centers obsolete.
Like all the major public cloud providers, Microsoft Azure doubled down on next-generation applications that rely on serverless computing and machine learning. Among the new products are Machine Learning Workbench, intended to improve productivity in developing and deploying AI applications, and Azure Event Grid, which helps route and filter events built in serverless architectures. Some important upgrades to Azure IoT Suite included managed services for analytics on data collected through connected devices, and Azure IoT Edge, which extends Azure functionality to connected devices.
Many of those Azure features are too advanced for most corporations that lack a team of data scientists. However, companies have begun to explore other services that rely on these underlying technologies in areas such as vision, language and speech recognition.
AvePoint, an independent software vendor in Jersey City, N.J., took note of the continued investment by Microsoft this past year in its Azure Cognitive Services, a turnkey set of tools to get better results from its applications.
“If you talk about business value that’s going to drive people to use the platform, it’s hard to find a more business-related need than helping people do things smartly,” said John Peluso, Microsoft regional director at AvePoint.
Microsoft also joined forces with AWS on Gluon, an open source, deep learning interface intended to simplify the use of machine learning models for developers. And the company added new machine types that incorporate GPUs for AI modeling.
Azure compute and storage get some love, too
Microsoft’s focus wasn’t solely on higher-level Azure services. In fact, the areas in which it caught up the most with AWS were in its core compute and storage capabilities.
The B-Series are the cheapest machines available on Azure and are designed for workloads that don’t always need great CPU performance, such as test and development or web servers. But more importantly, they provide an on-ramp to the platform for those who want to sample Azure services.
Another Azure feature addition was the M-Series machines that can support SAP workloads with up to 20 TBs of memory, a new bare-metal VM and the incorporation of Kubernetes into Azure’s container service.
“I don’t think anybody believes they are on par [with AWS] today, but they have momentum at scale and that’s important,” said Deepak Mohan, an analyst at IDC.
In storage, Managed Disks is a new Azure feature that handles storage resource provisioning as applications scale. Archive Storage provides a cheap option to house data as an alternative to Amazon Glacier, as well as a standard access model to manage data across all the storage tiers.
Reserved VM Instances emulate AWS’ popular Reserved Instances to provide significant cost-savings for advanced purchases and deeper discounts for customers that link the machines to their Windows Server licenses. Azure also added low-priority VMs– the equivalent to AWS Spot Instances — that can provide even further savings but should be limited to batch-type projects due to the fact that they can be pre-empted.
Jason McKaysenior vice president and CTO, Logicworks
The addition of Azure Availability Zones was a crucial update for mission-critical workloads that need high availability. It brings greater fault tolerance to the platform through the ability to spread workloads across regions and achieve a guaranteed 99.99% uptime.
“It looks to me like Azure is very much openly and shamelessly following the roadmap of AWS,” said Jason McKay, senior vice president and CTO at Logicworks, a cloud managed service provider in New York.
And that’s not a bad thing, because Microsoft has always been good at being a fast follower, McKay said. There’s a fair amount of parity in the service catalogs for Azure and AWS, though Azure’s design philosophy is a bit more tightly coupled between its services. That means potentially slightly less creativity, but more functionality out of the box compared to AWS, McKay said.
Databases and private data centers
Azure Database Migration Service has helped customers transition from their private data centers to Azure. Microsoft also added full compatibility between SQL Server and the fully managed Azure SQL database service.
Azure Cosmos DB, a fully managed NoSQL cloud database, may not see a wave of adoption any time soon, but has the potential to be an exciting new technology to manage databases on a global scale. And in Microsoft’s continued evolution to embrace open source technologies, the company added MySQL and PostgreSQL support to the Azure database lineup as well.
The company also improved management and monitoring, which incorporates tools from Microsoft’s acquisition of Cloudyn, as well as added security. Azure confidential computing encrypts data while in use, in addition to encryption options at rest and in transit, while Azure Policy added new governance capabilities to enforce corporate rules at scale.
Other important security upgrades include Azure App Service Isolated, which made it easier to install dedicated virtual networks in the platform-as-a-service layer. The Azure DDoS Protection service aims to protect against DDoS attacks, new capabilities put firewalls around data in Azure Storage, and end points within the Azure virtual network limit the exposure of data to the public internet to access various multi-tenant Azure services.
Azure Stack’s late arrival
Perhaps Microsoft’s biggest cloud product isn’t part of its public cloud. After two years of fanfare, Azure Stack finally went on sale in late 2017. It transfers many of the tools found on the Azure public cloud within private facilities, for customers that have higher regulatory demands or simply aren’t ready to vacate their data center.
“That’s a huge area of differentiation for Microsoft,” Mohan said. “Everybody wants true compatibility between services on premises and services in the cloud.”
Rather than build products that live on premises, AWS joined with VMware to build a bridge for customers that want their full VMware stack on AWS either for disaster recovery or extension of their data centers. Which approach will succeed depends on how protracted the shift to public cloud becomes — and a longer delay in that shift favors Azure Stack, Mohan said.
Trevor Jones is a senior news writer with SearchCloudComputing and SearchAWS. Contact him at firstname.lastname@example.org.