It’s been interesting to watch the big three public cloud providers select their on-premises dance partners over the past few years. The goal is to construct a platform that can stretch cloud based services into on-premises deployments into what is commonly being referred to as a hybrid cloud model. I fully admit, however, that the term hybrid cloud is nebulous in its definition (pun intended?) based on who you are talking to and what their perspective is focused upon within the stack.
I was recently invited to the annual Google Next 19 event and the invite-only Community Summit to hear more about what Google is doing in the realm of DevOps, SREs, and Hybrid Cloud. I’ll share some thoughts on Google Anthos and other hybrid cloud plays in this post.
Azure and AWS Hybrid Cloud Solutions
The first, Microsoft Azure, decided to tackle the hybrid cloud model with Azure Stack. This appliance “black box” instantiation of limited Azure services that have waterfalled down from the production instances of Azure is a combination of a hyper-converged infrastructure (HCI) stack with Azure services deployed. Administrative access is limited to the layers that live above the platform. We did a show entitled “Unpacking Azure Stack For Hybrid Clouds” on the Datanauts back in 2017 if you want to hear about the creation of this.
Next up is AWS. Their first dabbling into the on-premises world really came about with the introduction of AWS Snowball Edge, which I wrote about here in the post “Extending the Hybrid Cloud with AWS Snowball Edge.” It was the first strong signal for a move towards building hybrid clouds from a company that has traditionally scoffed at any deployments that live in the four walls of your data center. Now, we see that this pivot has completed with last year’s announcement of AWS Outposts which we covered a bit on the “AWS Outposts, Choosing An NVMe Fabric, Parallel NFS Cautions” episode of the Datanauts. This offering is a bit different in its stated architecture, but the end goals seem fairly similar to what Azure Stack is looking to do. Perhaps with a bit stronger of an opinion on how to do hardware? 🙂
Google’s Anthos Announcement
This week, Google has joined the club with the announcement of Anthos, an evolution of Cloud Services Platform, at the Google Next 19 conference in San Francisco. I suggest watching this interview that theCUBE did with Aparna Sinha & Chen Goldberg from Google that covers a bit on Anthos, as well as reading this extremely well pieced together site from Google. Their aim is to target the hot bed of innovation in the form of containerized applications that are tethered together with services meshes on Kubernetes in both cloud and on-premises environments. In fact, the day one keynote even showed an application running in AWS that was being controlled by Anthos, so this isn’t just another “let’s get people on GCP” play.
Considering that adoption of this platform is all about giving you, the user, the choice on how and where you want to run containerized applications, and it supports GKE on-premises, this gets two thumbs up from me. GKE on-premises is all about:
- Unified multi-cluster registration and upgrade management
- Centralized monitoring and logging with Prometheus and Stackdriver
- Hybrid identity and access management
- GCP Marketplace for Kubernetes applications
- Professional services and enterprise-grade support
I personally plan to dig into this further, especially as it relates to Serverless (one of my favorite technology focus areas at the moment and an inspiration behind Roxie on AWS Lex). Regardless of where you are focused on the greater hybrid cloud stack, it’s definitely interesting days ahead as all public cloud providers are now solving the mysteries of combining on-premises with public cloud in their different ways. Solutions that focus on removing toil and friction while increasing the user experience and overall velocity will end up as the winners.
