The operating models for enterprise IT are in a period of great flux, with cloud providers and business hardware and software vendors all positioning themselves for a favored spot in emerging infrastructure and application strategies. Transitional eras such as this create much confusion about system architectures, business structures, even terminology, which has led to a myriad of hyphenated cloud modifiers like “hybrid,” “multi” and “native.”
Most of these describe a future in which organizations operate on infrastructure from multiple service and software providers, each with a unique, but somehow interoperable implementation of a cloud software stack. The presumption behind all these hyphenated cloud models is that enterprises will use some cloud-independent management software to create a heterogenous, cloud-agnostic infrastructure and application service backbone that hosts next-generation enterprise systems. It all sounds hopelessly complicated.
As I detailed in last week’s column, AWS utterly rejected this conventional wisdom at it’s recent Re:Invent conference when the words “multi-cloud” or “hybrid cloud” didn’t once cross the lips of CEO Andy Jassy during his three-hour keynote. Instead, he described a series of new products and services that extend AWS services from its hyperscale data centers to smaller and more localized facilities owned by others, whether customers or network carriers.
By positioning AWS as the one superior cloud service foundation for next-generation enterprise applications, Jassy might have seemed arrogantly out of step with the IT executive and analyst party line that says complicated business systems require a multitude of software and service platforms, each optimized for a particular task and that converging application architecture to a single cloud stack is imprudent and unrealistic.
However, AWS isn’t alone with its strategy. While Jassy has loudly staked a claim for AWS as the cloud über alles, Microsoft has been furtively building a similar horizontally scalable cloud based on Azure and Azure Stack. To contrast with the heterogeneous multi-cloud concept of conventional wisdom, I call such homogeneous systems an omni-cloud.
Putting Microsoft’s Azure pieces together
Microsoft hasn’t been as publicly blunt as AWS in describing its one-cloud vision, but it has revealed enough pieces, many of which are in recent announcements, to see an omni-cloud strategy coming together. Microsoft CEO Satya Nadella provided clues as to the Azure-everywhere strategy when he answered a question about the company’s “hybrid engagements” at Microsoft’s latest Q1 2020 earnings call (emphasis added),
Overall, our approach has always been about this distributed computing fabric or thinking about hybrid as not as some transitory phase, but as a long-term vision for how computing will meet the real world needs because if you think about the long-term, compute will migrate to wherever data is getting generated and increasingly there will be data generated in the real world, where just when you think about the cloud, you have to think about the edge of the cloud as a very first-class construct. And so in that context, what we see is a couple of things that you see even in the results today.
… At our Ignite Conference, you will see us even take the next leap forward even in terms of how we think about the architecture inclusive of the application models, programming models on what distributed computing looks like going forward. So we feel well positioned there.
Note that Nadella echoes language that Azure executives have long used when discussing the cloud, that it is an operating and service model for infrastructure and applications, not a place. While they initially made this distinction to illustrate the seamless API consistency and Azure cloud integration of on-premises Azure Stack, as Nadella indicates, recent developments have seen Microsoft extend the Azure model to 5G mobile edge networks via Azure Network Edge (NEC) and multi-access edge (MEC), branch offices, retail sites and remote, environmentally harsh environments with Azure Stack Edge and embedded IoT devices with Azure IoT Edge. Sounds a lot like AWS’s cloud everywhere strategy, doesn’t it?
Extending Azure to the customer and mobile edge
Much like AWS is doing with Outpost, Local Zones, Wavelength, IoT Core and IoT Device Management, Microsoft is extending Azure services from its data centers to various locales including:
- On-premises private data centers with Azure Stack
- Branch offices via Azure Stack HCI
- Smaller remote locations, particularly those with many connected (IoT devices with Azure Stack Edge
- Mobile customer on-premises using Azure Stack Edge for mobile edge compute (MEC) and Azure Stack racks for network edge compute (NEC)
- IoT devices using Azure IoT Edge
- Localized low-latency access via network edge POPs and services like Azure Front Door (AFD, global network routing customization), ExpressRoute (high-speed private circuits linking customer and Azure infrastructure) and expansion of the ExpressRoute meet-me peering locations and edge sites with sub-30 millisecond latency.
Although Microsoft doesn’t have anything like the just-announced AWS Local Zones (micro-data centers with sub-5 millisecond latency to customers in a particular metropolitan area), if customers want them, such localized micro-regions won’t be hard to add given Microsoft’s expansive network.
Here’s how Azure’s senior marketing director describes the burgeoning Azure Stack portfolio of remote cloud devices (emphasis added):
Azure Stack portfolio is an extension of Azure to consistently build and run hybrid applications across datacenters, edge locations, remote offices, and cloud. Azure Stack provides customers choice and flexibility based on their solution needs from consistent hybrid cloud on-premises with Azure Stack Hub that can be connected or disconnected from public cloud, to high-performance virtualization on-premises with Azure Stack HCI or an Azure managed appliance that provides intelligent compute and AI at the edge with Azure Stack Edge.
Indeed, Azure Stack Edge will support VMs, Kubernetes clusters and deep learning acceleration using embedded GPUs. In contrast, Azure NEC is effectively an enterprise Azure Stack implementation tuned for mobile network operators, making it functionally similar to AWS Wavelength. Microsoft recently announced a partnership in which AT&T will use Azure NEC to run its SDN virtualized network services, aka the AT&T Network Cloud, in 5G edge sites (presumably base stations and the radio access network or RAN). Over the next five years, AT&T is also shifting to public-cloud infrastructure by migrating most non-network workloads to Azure.
Microsoft has pilot applications for its edge infrastructure in a variety of industries including logistics, retail analytics and fulfillment, smart agriculture, traffic engineering, drone tracking, gaming and AR.
In their totality, recent Microsoft efforts to extend Azure services outside its data centers to a variety of remote locations and device form factors illustrate a strategy almost indistinguishable from that AWS discussed at re:Invent. Indeed, one could substitute “Microsoft Azure” for “AWS” in the following quote from last week’s column with a few small exceptions as noted.
Re:Invent 2019 [Microsoft Ignite] shows an AWS unabashedly promoting [Microsoft delicately nudging customers and developers into] a cloud-native future with AWS [Azure] at its center. In this model, the AWS [Azure] ecosystem is the foundation for business applications, data and management controls, but one that is no longer hyperscale-centric, but instead can be delivered at the scale and location required by the workload. While the vision doesn’t banish corporate data centers to oblivion, it does render them secondary, as necessary appendages to the AWS [Azure] environment where required by law or regulation (data privacy) or performance.
While AWS’s Jassy has banished the terms “hybrid” and “multi-cloud” from his vocabulary, Microsoft still uses the hybrid term, although the future it describes is indistinguishable. Microsoft’s “distributed computing fabric” using a consistent set of Azure services, APIs and software development tools and that can be delivered not just from hyperscale data centers, but hardware as small as an embedded appliance or as large as multiple equipment racks, sounds just like Amazon’s AWS everywhere approach.
The key difference is in style, not substance. Where AWS brashly states its intention to boldly go where no cloud has gone before, Microsoft is diplomatically hand-holding its customers through what it knows to be a difficult technical and cultural transition. In either case, enterprises must pick a dominant cloud stack since intermixing them leads to needless inefficiency and management heartache.