Azure Stack is Microsoft’s hybrid cloud solution, an extension of Azure on off-the-shelf hardware that mixes cloud-like pricing and operations with the ability to keep your data in your data center. It’s a powerful tool, offering many of the same services you find in Azure, and adding more with each regular update. With Azure-consistent APIs code can migrate between cloud and data center, and a familiar set of tools for creating and managing resources, it’s become a bridge between regulatory compliance, restricted bandwidth, and the capabilities of a modern cloud platform.

But Azure Stack is no longer alone. Microsoft’s hyperscale-cloud competitors Google and Amazon are rolling out their own hybrid-cloud solutions, bringing their own hybrid cloud stacks to the market. It’s clear there’s a need for on-premises cloud-like systems, whether it’s to run workloads that can’t be run in the cloud due to regulatory restrictions, or where bandwidth is limited or expensive. It’s also a way to develop and test cloud code before deploying it globally.

SEE: The industry cloud: Why it’s next (ZDNet/TechRepublic special feature) | Download the free PDF version (TechRepublic)

There are other cloud-like architectures too: edge computing devices that provide gateways between on-premises resources like IoT devices and cloud services; or hyper-converged architectures that extend into the cloud for storage or additional compute resources when needed. Microsoft has solutions for both, in Azure Stack HCI and Azure Data Box Edge.

What then, of Google’s and Amazon’s offerings? They’re both currently building their alternatives to Azure Stack, in the shape of Google Cloud Anthos and AWS Outposts.

Introducing Google Cloud Anthos

Google Cloud’s Anthos is its Kubernetes-based take on how to deliver hybrid cloud services. Taking a big bet on containers seems to be working well for Google Cloud Platform, and Anthos takes much the same approach offering an on-premises version of GKE, the Google Kubernetes Engine. Code and container deployments built for GKE will run on your local Anthos installations, with support for the Istio Service Mesh and Google’s Cloud Run Knative-based serverless compute. A Migrate tool will help move existing virtual machines to containers, ready for use on GKE.

Anthos is purely a Kubernetes environment. That keeps you from running core GCP services on your Anthos installation. However, if you have sufficient bandwidth, the GCP Hybrid Connectivity tools will link Anthos to GCP and the rest of its services. Along with Kubernetes, Google provides tools like Apigee for API management, Cloud Build for application development, and Stackdriver for monitoring and management.

While it’s possible to set up and run Anthos on your own hardware (and even on Azure or AWS), Google is partnering with hardware vendors to deliver complete Anthos stacks. Several will be familiar, as they offer Azure Stack implementations. It’s not surprising, of course, as there aren’t that many server vendors and it would be difficult for any one software company to tie a vendor to an exclusivity deal. The same is true for Anthos’s services partners, which includes many of the same big business and IT consultancies as the Azure Stack program.

Introducing AWS Outposts

Another alternative is AWS Outposts. With a focus on infrastructure as a service, AWS Outposts offers two ways of deploying hybrid infrastructures in your data center. One option uses VMware Cloud to offer the same set of VMware tools as in your network and on Amazon’s servers, while the other hosts native EC2 features. With AWS Outposts, Amazon delivers a preconfigured server rack to your data center, running its own custom hardware. It’s not completely a black box; you have the option of choosing the compute, memory, and storage configuration to fit your application’s resource requirements.

The rationale for two different approaches makes sense. If you’re using VMware Cloud to host applications on AWS, then you can bring them down to a local AWS Outposts installation where there may be bandwidth issues with a connection to AWS. Similarly, the EC2 option brings AWS-native applications in to your network, with the same management tools as you use on AWS.

AWS treats an AWS Outposts installation as an extension of your AWS virtual infrastructure, linking it to your nearest AWS region. That allows you to connect local resources to AWS-hosted resources and services seamlessly, as if your application was running in AWS. There’s no difference in the programming model, with the same APIs on the Outposts hardware. Initially it offers only EC2 and EBS services, with other AWS services on the roadmap.

Like Azure Stack, AWS Outposts has two levels of management. At the top level there’s the control plane you use to manage your applications and services. Then there’s AWS’s own access to the hardware, which it uses to keep the underlying software and drivers up to date. You don’t need to patch servers or upgrade OSs, as it’s all managed by AWS behind the scenes, with the same no-touch experience as running your code in their data centers.

The hybrid future

It’s clear that there’s an industry-wide understanding that hybrid cloud solutions like Azure Stack are a key edge technology. We’ve seen Microsoft and its partners take Azure Stack down from a full rack to half a rack or even less, with its rugged implementations. So, it’s good to see competition from both Google and Amazon. While their solutions aren’t as capable as Azure Stack, supporting far fewer services and scenarios, there’s a lot of ambition to go further (especially in AWS Outposts), with a significant roadmap of planned services. But these new stacks still don’t offer the same breadth of platform services, and so don’t support as many ways of working.

Azure Stack may give Microsoft a head-start, but with new contenders like these Microsoft can’t rest on its laurels. It needs to push forward on delivering a deep set of Azure-compatible services on Azure Stack, with Azure Stack HCI an alternative for smaller businesses that want to take advantage of Microsoft’s on-premises virtualization and storage tooling. Adding Azure Data Box Edge to the family gives Microsoft a range of different on-premises tools that offer support for a wide range of different hybrid cloud scenarios, from compute to virtual infrastructures to IoT.

Azure keeps on stacking up the cloud features on-premises

How Linux took over everything, including Microsoft Azure

Power BI and Azure as the future of enterprise analytics

Microsoft Azure: A cheat sheet