Microsoft today announced a slew of new collaboration and productivity services to kick off its Ignite conference in Orlando, Florida. Azure took the spotlight, headlined by Azure Arc — which effectively enables Azure services anywhere — and a revamped data analytics offering and Azure Stack Edge.
Azure Arc brings Azure products and management to multiple clouds, edge devices, and datacenters on any infrastructure, with the goal of unifying orchestration and governance. Resources served through Arc look and feel like they’re practically native, and they’re accompanied by auditing, compliance, and role-based access and security settings designed to scale.
Azure corporate vice president Julia White framed Arc as the evolution of Azure Stack, Microsoft’s service that enables consistent cloud models deployable on-premises. She asserts it will confer benefits like seamless updates and deployment without compromising on scalability.
“Enterprises rely on a hybrid technology approach to take advantage of their on-premises investment and, at the same time, utilize cloud innovation,” White wrote in a blog post. “As more business operations and applications expand to include edge devices and multiple clouds, hybrid capabilities must enable apps to run seamlessly across on-premises, multi-cloud, and edge devices while providing consistent management and security across all distributed locations.”
Arc — which is available in preview starting today — supports hardware running Linux and Windows Server, as well as Kubernetes clusters (through Microsoft’s Azure Kubernetes Service). Developers can use its controls to build containerized apps that take advantage of the Azure tools of their choice (like Azure Resource Manager, Azure Shell, Azure Portal, API, and Azure Policy), while IT teams can launch and configure the apps using GitOps-based configuration management.
“Today, we take a significant leap forward to enable customers to move from just hybrid cloud to truly deliver innovation anywhere with Azure,” added White. “At Microsoft, we understand that hybrid cloud capabilities must evolve to enable innovation anywhere, while providing a seamless development, deployment, and on-going management experience.”
Azure Data Services (Arc) and Azure Arc-enabled API Management
Azure Data Services (Arc), which is a part of Arc, allows customers to run software like Azure SQL Database and Azure Database for PostgreSQL Hyperscale on most infrastructures. It eliminates the need to custom-code apps by fielding deployment from cloud to SQL database and by affording a choice in data, analytics engine, and resource type.
Azure Data Services (Arc) begins its preview November 4, starting with the ability to run Azure SQL Database and Azure Database for PostgreSQL Hyperscale on Kubernetes clusters.
Also debuting in preview is Azure Arc-enabled API Management, a containerized API gateway that allows users to manage APIs within the Azure API Management plane. It can be deployed in any environment and keeps data localized for security and compliance purposes.
Azure Stack Edge
On the subject of Azure Stack — which Microsoft says is actively used by enterprises across 60 countries, including the Hong Kong Stock Exchange, KPMG Norway, and Airbus Defense & Space — today marks the launch of Azure Stack Edge, a managed AI-enabled appliance designed to bring compute, storage, and intelligence to the edge. In the near future, Azure customers will be able to take advantage of features like virtual machine support and graphics chip acceleration, in addition to multiple nodes and multi-access compute.
The above-mentioned appliance doubles as a storage gateway and comes with a built-in FPGA that accelerates AI inferencing tasks. Customers will be able to run apps on containers and take advantage of a fail-over feature that moves virtual machines to other servers if needed.
Azure Stack Edge will arrive alongside a new Rugged series of Azure Stack machines intended to bring cloud computing capabilities to “the harshest environment conditions.” Two new machines will target defense, industrial, energy, humanitarian aid, and emergency response scenarios.
Azure Stack Hub
Kubernetes is now generally available on Azure Stack Hub, Microsoft’s extension that lets customers deliver Azure services from their own datacenter or consume them from a service provider. Clusters contain a Kubernetes Cloud Provider that leverages Azure Resource Manager to scale compute, network, and storage through an Azure-maintained virtual machine image.
Windows Virtual Desktop on Stack Hub is launching in preview, enabling Host Pools to run workloads on Azure Stack Hub. As for BC/DR foundational pattern, which will be available in the first half of 2020, it’ll provide virtual machine failback templates for distributed servers.
Event Hubs is launching in preview. It’s a real-time data ingestion product that can be used to stream millions of events per second and build dynamic data pipelines. So is Stream Analytics, which lets users design hybrid, low-latency architectures on Stack Hub that surface data streams from IoT Edge or Event Hubs and output the results to local services and databases.
Azure Bastion, a fully managed platform-as-a-service offering that provides RDP and SSH access (using SSL) to virtual machines through the Azure Portal, is now generally available in six Azure regions: Australia East, East U.S., Japan East, South Central U.S., West Europe, and West U.S. Sessions can be initiated directly from Azure Portal (using single click), after which Bastion brokers a session to target machines in the network over private IP addresses.
Azure Disks, Microsoft’s disk storage offering for use with Azure virtual machines, now boast better performance (in preview) for workloads with less predictable traffic patterns, courtesy of bursting enhancement. Applicable SSD disks can now burst up to 30 times the provisioned performance target.
Disks are available in smaller sizes — 4GB, 8GB, and 16GB — and now support server-side encryption with customer-managed keys (SSE with CMK) in Azure Key Vault, in preview for Premium SSD, Standard SSD, and Standard HDD disk types. Furthermore, Azure Managed Disks now offer direct upload of virtual hard disks from on-premises or backup locations as managed disks.
New virtual machine types
Azure Da_v4 and Das_v4 virtual machines, which launched in preview last August, are now generally available and deployable via the Azure Portal, command-line interface, or PowerShell, or using Arm Templates. They pack AMD Epyc 7452 processors and provide up to 96 virtual processors, paired with 384GBs of RAM and 2,400GBs of SSD-based temporary storage.
The new virtual machines join Ea_v4 and Eas_v4 instances, which provide up to 96 virtual processors, 672 GBs of RAM, and 2,400 GBs of SSD-based temporary storage, and Azure Generation 2 virtual machines. Generation 2 boasts increased memory and Intel Software Guard Extensions, plus support for the UEFI boot architecture and operating system disk sizes exceeding 2TBs.
HBv2 virtual machines are on the way and will enable customers to run a single job across 36,000 cores with support up to 80,000 cores. And generally available starting today is the HPC Cache service, which automatically caches data both on-premises and in the cloud by delivering network file system (NFSv3) access in compliant directory structures.
Last, but not least, are additions to Azure’s NV size virtual machines family optimized for design, gaming, machine learning, and simulation workloads.
The NDv2-series features eight Nvidia Tesla V100 interconnected graphics chips with 32GB of memory each, 40 non-hyperthreaded Intel Xeon Platinum 8168 processor cores, and 672GB of system memory, as well as 100Gbps EDR InfiniBand with support for standard Mellanox OFED drivers and all message passing interface (MSI) types and versions. As for NVv4 — which is available for preview starting today — it boasts a total of 256GB of graphics chip memory and AMD’s MxGPU, plus SR-IOV technology, which offers a dedicated graphics frame buffer.
In a related development, new virtual machine scale sets launched in preview. They let customers create an empty scale set and add various virtual machines belonging to different series later, and to provision virtual machines with custom images using a shared image gallery. A complementary new scale-in policy, instance protection from scale-in, and termination notifications help determine the order in which virtual machines should be scaled and provide up to 15 minutes to perform cleanup or end tasks prior to deprovisioning.