Google Cloud Platform News at Cloud Next '18

Kevin Krause (left) with his Ubilabs colleagues in front of the San Francisco skyline. Cloud Next '18 took place at Moscone Center in San Francisco. One of our conference highlights was the announcement of BigQuery GIS. Kevin Krause enjoying the after-hours program at Cloud Next ‘18.

Cloud Next '18 was all about the Google Cloud Platform – an extremely secure yet also very user-friendly platform. Ubilabs' Cloud specialist Kevin Krause attended the conference in San Francisco and summed up the most interesting GCP innovations.

1. Best practices for web applications

My personal highlight is my signed first edition copy of the book "Site Reliability Engineering", which was presented exclusively at Cloud Next. The authors describe how to set up a web application so that it runs as smoothly and reliably as possible and provide hands-on best practices.

2. 100% availability shouldn’t be the goal

In their talk "Improving Reliability with Error Budgets, Metrics, and Tracing in Stackdriver", Marie Cosgrove-Davis and Liz Fong-Jones showed how to increase the reliability of web applications. However, they also clearly argued that 100% availability of web applications is neither a reasonable nor a realistic goal.

3. Safety first

I was thrilled hearing about the many new security features that were presented in the talks: hardware for managing crypto keys, hardened virtual machines running only signed code, and Google's new "Titan" hardware dongles for 2-factor authentication to name a few. These features further increase the reliability of the Google Cloud Platform and make it even easier for companies to set up robust security measures.

4. TPUs for local machine learning

In one of the keynotes, Google presented TPUs that you can install in your own devices, enabling efficient machine learning on site. The Tensorflow chips can be integrated into tablets and smartphones or even devices such as traffic cameras and can then be used for mobile machine learning without cloud or internet.

5. How to design good APIs

Google once again shows how it’s done: In his talk "Designing Quality APIs", Martin Nally explained how to build APIs that are not only easy to understand and easy to use, but that are also robust – a very good hands-on talk.

6. BigQuery GIS: Public alpha

Another highlight was the announcement of the public alpha version of BigQuery GIS. Not only can you use the GIS functionalities for huge amounts of data and analyze geo data in the cloud, you can also make predictions using machine learning in BigQuery.

Interested in a deep dive? Watch our favorite talks in our Google Cloud Next playlist on YouTube.