At the Google Cloud Platform developer conference today Google explained its vision of how enterprises will move to and operate more efficiently in its cloud. The vision: Zero-Dev-Ops cloud computing that “instead of programming a computer you teach it what it [Google’s cloud] wants to know and it learns to give you what you want,” according to Alphabet Chairman Eric Schmidt.
The vision may sound more like a SciFi movie, but it knit together new and recently announced technologies into a coherent explanation.
Google plans to bring its enormous-scale data center know-how learned fulfilling 85% of the world’s searches with open source software in a serverless secure cloud in which it invested $10 billion in capital equipment last year. It has repurposed open source versions of its highly optimized internal systems used to build and operate the seven services such as Maps and Gmail that have more than a billion users.
Google CEO Sundar Pichai, Cloud Executive Vice President Diane Greene and Alphabet Chairman led with the company’s robust security, claimed its provisioning and use of cloud resources was portable and more efficient and less expensive than Amazon’s, stated a commitment to reduce its customers cost of managing virtualized infrastructure called dev-ops to nearly nothing and emphasized open source software over its cloud products.
The daily reports of corporate data breaches is the reason security was discussed first. Google has the security scale, technical expertise, tools and analytics built into its infrastructure that enterprises and their vendors can’t reproduce. If there were just one reason an enterprise should look at moving to the cloud it is security.
Google claimed that its provisioning efficiency only uses the resources needed to meet the workload, neither over-provisioned individual servers nor a contractually provisioned minimum number of servers. Greene claimed 50% better price performance using Google’s cloud.
Snapchat growth from 0 to 100 million users without a dedicated dev-ops team was cited as an example of dev-op cost reduction. This presupposes using open source containers and orchestration software based on what Google uses internally. An enterprise moving virtualized server workloads to Google’s cloud couldn’t achieve these gains without restructuring and rewriting systems.
Google spoke extensively about open-source Docker containers and Kubernetes orchestration software that is an operational advantage and a tactic for competing with market-leader Amazon. Docker containerizes applications with all the libraries and dependencies an application needs to run. Containers have been a Linux feature that was popularized by Docker, the company with the same name. Docker container use by Google’s cloud customers is doubling every quarter because they have the advantage of portability and serve as a common unit for test and for continuous deployment.
+ ALSO ON NETWORK WORLD Google Graveyard: What Google has killed off in 2015 +
Kubernetes orchestrates the workload of Docker containers in response to the demand. Designed to manage thousands of Docker instances, Kubernetes can spread a workload across a hybrid cloud using on-premise equipment, Google’s and Amazon’s cloud services. Google engineer Ray Tsang demonstrated an impressive feature called rolling-updates through the open source Stackdriver dashboard announced today. Tsang launched an update that replaced Docker instances with new versions and using load balancing that will spread the workload between the Google Cloud and a bare metal Intel based server rack with Kubernetes.
The open source machine learning project Tensorflow and its applications was talked about and demonstrated frequently. A product of Google’s long term investment in artificial intelligence, it is a differentiator that will attract customers to its cloud who need machine learning. Available as an application programming interface (API) within Google’s cloud, Tensorflow is used to interpret digital and non-digital information such as images, video and voice by teaching it with training data sets. Google Photos was used as an example of how Tensorflow could not only sort and identify subjects with facial recognition but classify them based on activity.
Google has a challenge with new companies and enterprises though. Google’s open source, vendor independent cloud with low dev-ops demands should be an attraction to new companies building state of the art applications. Google’s marquee customers Spotify and Snapchat are impressive but Amazon has a much more extensive list including Netflix, Airbnb and Uber. Google’s challenge in attracting the developers in these agile small companies who already know Amazon’s cloud is to grab their attention then convince them to take the time to evaluate and learn their offering.
Google clearly understands its challenge with enterprises. Moving legacy systems to the cloud doesn’t have the same return or strategic advantages as apps built anew by more agile small companies. Green said that Google would become an applied R&D team to give their enterprise customers a boost.