It’s no surprise that top cloud providers, Amazon Web Services, Microsoft Azure and Google, continued to dominate technology headlines in 2017. This year, we saw these cloud giants perform the same one-upmanship around tools, services and prices that we have in the past — but this time, with a sharper focus on technologies such as containers and hybrid cloud.
Before you head into 2018, refresh your memory of SearchCloudComputing’s top news from the past year:
Amazon, Microsoft crave more machine learning in the cloud
All the top cloud providers see the importance in machine learning, and Amazon Web Services and Microsoft Azure put their differences aside in October to jointly create Gluon, an open source deep learning interface based on Apache MXNet. This new library is intended to make AI technologies more accessible to developers and help them more easily create machine learning models. In the future, Gluon will work worth Microsoft Cognitive Toolkit.
Meanwhile, Google Cloud Platform offers TensorFlow, another open source library for machine learning. While TensorFlow is a formidable opponent, some developers shy away from it due to its complexities.
The main problem that all providers face in this space is that the public cloud isn’t always the best environment for complex machine learning workloads due to cost, data gravity or a lack of skill. Some data scientists continue to use the public cloud to test, but then run the workloads on premises.
Google hybrid cloud strategy crystallizes with Nutanix deal
While cloud is popular, many workloads are still kept on premises — either due to their design or compliance issues. Top cloud providers continue to seek partnerships to target the hybrid market and ease the gap between data centers and the cloud.
The Amazon and VMware deal tends to be the most common example of this. But in June 2017, Google partnered with Nutanix to fuel its own hybrid efforts. Next year, customers will be able to manage and deploy workloads between the Google public cloud and their own hyper-converged infrastructure from a single interface. This partnership will also extends Google cloud services, such as BigQuery, to Nutanix customers, and enable customers to use Nutanix boxes as edge devices.
Kubernetes on Azure hints at hybrid cloud endgame
One of containers’ main advantages is enhanced portability between cloud platforms — a feature that’s especially attractive to hybrid cloud users. In February 2017, Microsoft unveiled the general availability of Kubernetes on Azure Container Service (AKS, formerly ACS), making it the first public cloud provider to support all the major container orchestration engines: Kubernetes, Mesosphere’s DC/O and Docker Swarm.
The move was one that could especially benefit hybrid cloud users because both Docker Swarm and Kubernetes enable teams to manage containers that run on multiple platforms from a single location. In October, Azure rolled out a new managed Kubernetes service, and rebranded ACS as AKS. AWS countered in November with Amazon Elastic Container Service for Kubernetes, a managed service.
Azure migration takes hostile approach to lure VMware apps
To compete with VMware Cloud on AWS, Microsoft released a similar service for Azure in November 2017 — without VMware support.
Azure Migrate enables enterprises to analyze their on-premises environment, discover dependencies and more easily migrate VMware workloads into the Azure public cloud. A bare-metal subset of the service, VMware virtualization on Azure, is expected to be available in 2018, and enables users to run a VMware stack on top of Azure hardware. While the service is based on a partnership with unnamed VMware partners, and involves VMware-certified hardware, the development of it didn’t directly involve VMware itself, and cuts the vendor out of potential revenues. VMware has since said that it will not recommend or support the product.
Cloud pricing models reignite IaaS provider feud
The price war continued in 2017, but top cloud providers changed their tune: instead of direct cuts, they altered their pricing models. AWS abandoned its per-hour billing, in favor of per-second billing, to counter per-minute billing from Google and Azure. Google shortly responded with its own shift to a per-second billing model.
Microsoft, for its part, added a Reserved VM Instances option to Azure, which provides discounts to customers that purchase compute capacity in advance for a one- or three-year period. The move was a most direct shot at AWS’ Elastic Compute Cloud Reserved Instances, which follow a similar model.