Think Less with Google Cloud OnBoard

Google is making it easier for developers to simply think less and build more.

Oct 4, 2017 · 4 min read

In the late 1990s, Apple unveiled a new marketing slogan, “Think Different.” It was meant to evoke ideas about non-conformity and creativity in an age of beige business PCs typically used for word processing and financial calculations.

After attending the Google Cloud OnBoard event in Washington DC today to learn about the capabilities and performance of Google’s Cloud Platform, one might think Google could bring back that old Apple slogan to once again make people think about what’s possible — this time in cloud computing — rather than what’s already been done. Alternatively, perhaps Google could play with that old slogan a bit, and just say “Think Less.”

As we move into a world soon to be dominated by connected devices, technology professionals face increased loads of complexity. Online services won’t just be exercised by users, but by things. Everywhere. All the time. And anyone who has ever spent time on pager duty for a technology company knows that means bigger workloads and higher levels of mental strain across teams.

Google, with their decades of experience working at “Google-scale,” have not just created a cloud platform that can compete with anyone on price and performance, but one loaded with features meant to reduce strain for their Cloud Platform customers.

The Cloud Platform is full of little “developer ergonomics,” that help people worry more about their own software and less about operational concerns. For example:

Any time you create a Virtual Machine under a network, it’s instantly able to communicate via internal IP with every other machine on the network, regardless of region. For IoT service providers, this is extremely helpful. Imagine your service requires that all external-facing servers can communicate with a centralized cluster somewhere in North America, but your IoT service has also recently seen a big spike in popularity in Asia. This is a classic scenario where success could be breaking things. But thanks to the design decisions made for Google Cloud Platform, it’s easy to deploy new, Asia-based VMs that can immediately communicate with every other machine in your network regardless of geographic region.
Google App Engine, Google’s original Cloud Service, may well have been a great idea ahead of its time. Today, Google App Engine is offered in two forms: Standard and Flexible. Google Standard makes it possible to deploy entire services with immediate elastic scalability with just a couple of lines of code. And intuitive, web-based deployment tools makes deployment of new versions of your app (including rollbacks, A/B testing, and service sunsetting) relatively simple. You can even entirely “shut down” a web service for all of your userbase except for that one extremely important customer who just can’t handle the breaking changes your new version introduces, but has purchased more of your Enterprise IoT devices than anyone else. Them? Well, they get a simple alternative URL to reach their beloved, archaic version of your service (and you get to keep them as a customer).
Few things have had a bigger effect on “developer ergonomics” and continuous deployment of services at scale than Docker. And nothing is more popular for orchestrating Docker-ized services than Kubernetes. Google Container Engine speaks Kubernetes natively. Other cloud platforms also support Kubernetes — that’s what makes Kubernetes so useful — but Google also happened to invent it.And Google Container Engine’s SDK and command line tools are built to provide Kubernetes with all the information it needs to easily deploy Docker container clusters (or pods) onto Google Cloud Platform.
Google Cloud invented a new kind of relational database that offers strong, global data consistency and horizontal scalability. For an IoT service provider, this could be huge. IoT service providers may have devices deployed all over the world that need exactly the same answer any time any of them request data from the service. Google Spanner makes it possible to provide that answer. And now Google’s customers don’t have to solve that problem themselves.

We are now in the early stages of an all-out market-share war between the major providers of cloud platforms. Many believe that this war will be fought largely on price. In fact, Amazon just this week announced new per-second billing on instances of VMs on EC2 — a move designed to lower costs for customers. But Google is making strides in competing not only on cost, but also in making it easier for developers to simply think less and build more.