Google LLC has begun to pick up steam in the cloud wars, though it still lags considerably behind Amazon Web Services Inc. and Microsoft Corp.’s Azure in market share in the public cloud.

One approach for gaining traction among enterprise cloud users is to build a substantial developer ecosystem. Today at Google Cloud Next ’18 in San Francisco, the company made several announcements that demonstrate its commitment to enabling a new generation of developers to build intelligent, production-ready applications for the growing range of enterprise hybrid and multicloud environments.

Building the cloud app-dev ecosystem

Google’s core go-to-market strategy in the cloud wars is to deploy a comprehensive application development environment on top of Kubernetes. Though AWS, Microsoft and most other cloud providers are committed to Kubernetes, Google has taken this open-source platform further strategically.

Google Kubernetes Engine is its core application server platform in the cloud. Within the core GKE-centric Cloud Services Platform announcement at today’s day-one keynote, the most significant new cloud initiatives were geared toward delivering “build once, run anywhere” Kubernetes applications into hybrid and other multicloud environments:

• On-premises: Google announced the imminent alpha of on-premises version of GKE, deliverable as a Google-configured and fully managed Kubernetes image from the cloud. The company said customers will be able deploy GKE On-Prem — which has the same behavior, look, and feel as the public cloud GCP version — in the private clusters of their choice and orchestrate these instances with GKE running anywhere in the public GCP.
• Multicloud: Google announced the general availability of a production-ready version 1.0 of its multi-cloud integration mesh and abstraction layer, Istio, which it launched a year ago, and which comes natively integrated with GKE. The vendor also announced that a fully managed GCP version of Istio will be going into alpha soon.
• Serverless: Google announced the GKE Serverless add-on for running serverless workloads anywhere on GKE clusters. It also announced a new open-source technology called “Knative” for deploying containerized apps and exposing them through serverless interfaces anywhere on GKE clusters.

To enable developers to build production-grade Kubernetes applications for deployment anywhere in public, private, hybrid and other multiclouds, Google made several key announcements today at Next ’18:

• Policy, identity and access management: Google announced that its new GKE Policy Management solution will be going into alpha soon. This fully managed solution will support centralized policy, identity and role-based access management anywhere on GKE clusters and across Istio.
• Service monitoring: Google announced its Stackdriver Service Monitoring offering now provides a visual dashboard for discovering, monitoring and analyzing service-level metrics across public and private GKE clusters over Istio.
• DevOps: Google announced the new CloudBuild, which provides a fully managed, end-to-end continuous integration and continuous delivery cloud service for building, operationalizing and managing cloud microservice apps anywhere on GKE clusters across Istio.
• API management: Google announced that its Apigee solution can now manage APIs anywhere on GKE clusters and in the Istio mesh.

The company also underlined the importance of Google Cloud Platform Marketplace — recently rebranded from its former name of Cloud Launcher — as its strategic channel for partners to deliver their Kubernetes applications to market.

Going forward, Wikibon would like to hear a bit more about how Google’s new Jib open-source tool, which is designed to simplify the task of containerizing Java apps, fits into its GKE app-dev toolchain. Today at Next ’18, there wasn’t any specific discussion of how Java developers will be able to easily containerize their apps for deployment and orchestration on GKE or Istio in multiclouds. If Google were to provide that sort of capability in the near future, they could win over a lot of established Java developers to their Cloud Service Platform and perhaps gain momentum vis-à-vis AWS in the public cloud and Microsoft in the hybrid public/private cloud.

Accelerating AI application development

Google already made its most significant AI announcements of the year so far a few months ago at both Google I/O and the TensorFlow Development Summit.

On Day One of Next ’18, Google’s chief AI-related announcements focused on accelerating the app-dev pipeline for machine learning, deep learning, natural language processing and other AI apps on the cloud and at the edge.

Key to these announcements were the automation of AI development tasks. This is absolutely essential in a world where people with these skills are few and far between.

Most significantly, Google announced the public beta of Cloud AutoML Vision, an AI app-dev automation service that has been in alpha testing since earlier this year, and in which about 18,000 Google customers participated. Image recognition is a hot AI app, especially for facial recognition, so Wikibon is not surprised that Google is prioritizing this use case in its go-to-market strategy for its AI app-dev pipeline automation platform.

Likewise, it’s no surprise that Google is rolling out new app-dev automation tools, as well as APIs, for natural language processing and machine translation. These are especially critical for building and refining the AI-powered conversational UIs that are remaking apps everywhere, such as those that drive contact-center agent productivity and customer satisfaction.

However, today’s AI-related app-dev announcements at Google Next ’18 felt a bit underwhelming. The vendor missed an opportunity to show progress in building developer-friendly tooling for the next generation of browser-based machine learning JavaScript apps built on its TensorFlow.js framework, or any mobile, embedded or edge apps built on TensorFlow Lite.

In hardware-related news, Google announced the alpha of the third generation of its cloud-based Tensor Processing Units or TPUs. However, it failed to provide any detail on how well these neural-network processing units, deployed in the Google Cloud Platform and available for developers’ AI training and inferencing workloads, stack up performance-wise against the second generation TPUs. It also failed to provide competitive performance benchmarks with respect to graphic processing units, third-party neural network processing units, field-programmable gate arrays, application specific integrated chipsets, and other neuromorphic hardware architectures in this hotly competitive segment.

Boosting knowledge worker productivity

Google devoted little attention to its office productivity suite at its recent I/O conference. However,G Suite  was a key focus of its keynote today at Next.

Google is clearly positioning G Suite as its encroachment strategy versus Microsoft in the latter’s core enterprise knowledge worker customer stronghold. All of the G Suite enhancements that were announced today are to assure that business customers that, if they wish, migrating to Google’s suite is a low-risk proposition. New features such as centralized threat mitigation, regional data copies and AI-augmented messaging, personalization and grammar correction will bring Google’s suite into more enterprise users’ comfort zones and keep Microsoft’s Office 365 from running away with this segment.

Nevertheless, Wikibon is a bit disappointed that there was no specific tie-in of the latest G Suite enhancements with the mobile productivity features announced at Google I/O. Business productivity is intimately associated with mobility, but there was no mention today of how the Android P predictive experience, Google Duplex natural language interface or Google Maps real-time navigation recommendations will be integrated with G Suite going forward. Introducing voice commands into Hangouts Meet hardware felt like more of a strategic distraction for Google, rather than a category-killer feature.

Also, Google missed a golden opportunity to assure business users that third-party developers would not be able to scan their Gmail going forward. Even though Google has stopped this practice in its own app development teams, the vendor’s vast AI partner ecosystem — which includes many developers that tap into Gmail and other user data the Google Cloud ecosystem — could easily deter further enterprise adoption of these productivity tool if the practice continues unabated.

For further discussion of these announcements, check out today’s interviews on theCUBE at Google Cloud Next ’18.

Photo: Robert Hof/SiliconANGLE