Microsoft Edge Zones brings connectivity to the fore, Azure-based compute to the edge
Microsoft has been making some important announcements over the last few months as it continues to flesh out its edge and 5G strategy. Following on its purchase of Affirmed Networks, it is worth closely examining the edge network services offered by Microsoft, called Azure Edge Zones. Whereas Affirmed Network’s target is the carrier market and particularly the mobile operator market, Azure Edge Zones are aimed at enterprises looking for a 5G or edge solution.
The following company briefing has been made available through our partnership with Structure Research, an independent research and consulting firm with a focus on the cloud, edge, and data center markets.
Details: Azure Edge Zones are small footprint versions of Azure which aim to bring data and applications closer to the end-user across deployment models and locations. It comes in three different offerings.
– Edge Zones: This is a wholly Microsoft owned and managed service that runs from Microsoft’s edge sites up to new edge sites, located in cages at exchange points, and have hosted network-centric products such as ExpressRoute, CDN, or edge caching and application acceleration. By adding Azure, Microsoft takes that one step further, bringing compute and storage closer to the end-user. The first three locations for the service are New York, Los Angeles, and Miami—densely populated areas but often at a distance from the cloud hubs located in places like Northern Virginia. Microsoft hopes to attract applications with low latency and intense processing requirements such as rendering or gaming. At present the service is rudimentary, consisting of just basic VMs, containers, and networking. For analytics or more advanced tools, customers have to push to the nearest Azure region.
– Carrier Edge Zones: This is essentially Azure integrated into a carrier network. Microsoft has been trialing this with AT&T for nearly a year. Mobile networks are traditionally designed to carry voice traffic and have highly centralized structures with few interconnect points. As data traffic has grown, this architecture has become a bigger and bigger issue: data is frequently backhauled to a central interconnect before being directed to the edge once more, in a process known as ‘tromboning.” Microsoft hopes to eliminate or at least mitigate this issue with edge nodes that will serve data locally.
– Private Edge Zones: This is an on-premise, small footprint version of Azure. It is based on Azure Stack Edge (Microsoft’s competitor to Amazon Snowball). The connectivity piece, again, is key. Microsoft has tightly integrated the product with SD-WAN services, greatly simplifying the deployment of applications across multiple hybrid environments. Microsoft expects to see Private Edge Zones on assembly lines in factories, where you might find, say, a private Azure Stack Edge box attached to an industrial camera, checking beer bottles for contamination. Assembly lines of course are a widely touted 5G edge case. Stack Edge is a physical device, so it seems that the competitor Microsoft has in mind here is Amazon Outposts.
Partnering: Microsoft already has a partner to help it take Private Edge Zones to market. Vodafone’s business division in the UK is to resell the product. This is interesting because Vodafone has a 5G partnership with Google. The Vodafone/Google platform though, dubbed Project Neuron, is a Big Data analytics platform that operates at the core generating the kind of insights necessary to manage a 5G network while the Microsoft collaboration is a sales proposition that sits at the network edge.
Context: How does Edge Zones fit into the rest of the Azure family? With a fair bit of complexity, it must be said. Microsoft has a long history of confusing naming conventions and its Azure family upholds the tradition. It re-branded and re-booted its edge and on-premise offerings last year as Amazon and Google started to show increasing interest in the area. Currently, we have these edge appliances and software in addition to Edge Zones:
– Azure Stack Hub: private cloud software, available on a range of hardware OEMs; originally based on Hyper-V and seen as a VMware competitor and then subsequently called Azure Stack, it has been on the market since 2017.
– Azure Stack Edge: Microsoft’s answer to Amazon Snowball, a 1U mountable appliance mainly designed to transfer data to Azure data centers but which can also perform some compute and storage functions; it was known as Azure Data Box until late last year. Note that it is this device, rather than Stack Hub, that is used for Private Edge Zones.
– Azure Stack HCI: a re-brand of Windows Server 2019, one of Microsoft’s long-running warhorses, used widely in larger enterprises.
– Azure Arc: extends Azure Resource Manager’s capabilities to Linux, Windows servers, Kubernetes containers across on-premise, multi-cloud, and edge.
Analysis: We have noted previously how Microsoft has taken a software approach to the cloud and edge. With Azure Edge Zones, it is connectivity that is taking the central stage. While that may be inevitable given that the product is aimed at 5G use cases, Microsoft is also drawing on its deep network knowledge. It is not widely known in the industry, but Microsoft is something of a network pioneer. Its widely open sourced switch operating system, SONiC, is the basis of its Azure network, which is flexible and easily configured. Microsoft wants Azure and Azure Edge zones to be the same: the goal is to enable enterprises to seamlessly deploy containers or VMs just about anywhere: on or off-premise, in the cloud, or in a factory. It is a bold undertaking and truly impressive if becomes reality.
Daniel Beazer, Contributing Analyst, Structure Research
Australian firm adds funding for edge data center deployments
Summit gathering explores edge computing, AI for vision
Article Topics
5G | Affirmed Networks | Amazon | Amazon Web Services | AWS | connectivity | edge | interconnection | Microsoft | Microsoft Azure | Structure Research | telco | Vodafone
Comments