14 Mar 2017 Cloud Next 2017 – Shifting to the Cloud
Last week I had the privilege of attending Google Cloud Next in San Francisco. With Google finally due to open a datacenter in Australia this year, it was certain to be a great opportunity to learn about what’s next with Google Cloud.
From the moment I arrived at the baggage carousel at San Francisco International Airport, I was swamped with advertising for the conference. It was clear that Google is really pushing their cloud platform to as many developers as possible. This left me really excited for what was about to come over the following week. In this post I’m going to try and sum up how it all went.
The first stop on my Cloud Next journey was a bootcamp. I choose to attend the Security Bootcamp where I spent the day learning about best practices for securing Google Cloud Platform resources. They covered topics such as IAM, VPC’s, VM management, security keys including best practices with all of these.
It was delivered using a mixture of presentations and hands-on labs where we were able to complete a variety of exercises using QwikLabs and running demos on GCP. Overall the exercises were easy to follow and designed well. Making use of GCP managed keys and automatically SSH-ing onto an instance with the click of the button made it easy to run through tutorials.
As part of Shine’s continued working relationship with Google, we were invited to the partner summit. It was clear that the partner summit was all about selling us the idea of moving to the cloud. When GCP started back in 2008 with the release of AppEngine they asked their customers to make their applications fit their vision of the cloud. Now fast forward to 2017 and this has changed.
After the partner summit, it seems clear that they are becoming more and more committed to making it easier for customers to move their existing on-premises applications to the cloud. They still have a lot of work to go but they are also expanding their focus to the enterprise market and not just focusing all their efforts on the Startup market.
Google Cloud Next is the largest and most high-profile conference that I have ever attended. I had high hopes and was really excited for the keynote on day 1. The keynote was opened by Dianne Green. Dianne started by explaining Google’s commitment to meeting customer needs for both enterprise applications as well as startup applications.
There was a noticeably strong focus throughout the keynote on moving enterprise applications into the cloud. The buzzword that was thrown around a lot was ‘lift and shift’, the process of lifting on-prem applications and shifting them into the cloud. I was a bit disappointed to see that there were no major announcements made during the keynote.
Instead of spending time on demoing a cool new product, we instead got to hear about SAP, Colgate-Palmolive and Verizon’s ambitions and goals for moving to Google Cloud. Dianne was doing the majority of the presenting throughout the keynote. With an often nervous laugh throughout the keynote, Dianne could have done a much better job presenting (not that I’d do any better in front of 10,000 people).
“The cloud with the best technology is the best cloud” – Dianne Greene
The best part of the keynote was presented by Fe-Fe Li, who is an AI researcher at Google. Emphasising the importance of Machine learning at Google, Fe-Fe went on to announce the general availability of the video intelligence API. Then Sara Robinson was introduced by Fe-Fe, who did a great job demoing the Video API.
Overall I was disappointed with the keynote and that was evident from a sizeable portion of people walking out while Dianne was summing up. Not keeping things focussed on developers – who formed a majority of the attendees – was a bad move in my opinion.
The good news is that day 1 was not all a let-down. I enjoyed all of the breakout sessions following the keynote and in particular the session I attended on Kubernetes. It was presented by Robert Bailey and Chen Goldberg, both engineers from Google. They presented using a mixture of slides and demos, showing how easy it is to get setup with Kubernetes and Google Container Engine. They also announced a new Auto Repair / Upgrade feature which is now in Beta. This allows the automatic upgrade of versions of Kubernetes, all in a reliable manner.
The day 2 keynote was leaps and bounds ahead of day 1. Urs Hölzle opened the day’s keynote with a talk on Google’s commitment to performance, reliability and security.
On the reliability front, Urs emphasised the fact that GCP customers are now serving over 1 Billion users every day from their highly secure cloud. It was also announced that there will be 3 new regions being added to the roadmap, taking the total number of regions worldwide to 17.
The best part of the day’s keynote was the demo of Cloud Spanner. Released only a couple of weeks ago, Cloud Spanner is a horizontal scaling relational database. It is being promoted as a globally-distributed database, however, at this stage, it is only capable of running within a single region. Multiregional support is set to come later in the year.
I went to the breakout session on Cloud Spanner and for me, it was defiantly the highlight of the conference. Presented by Dominic Preuss from Google and Peter Bakkum from Quizlet, it was a really engaging talk which introduced best practices on using Spanner as well as an insight into the inner workings of the database. One thing that was particularly talked-about was the concept of interleave tables, the most efficient way that Spanner can handle one-to-many table relationships.
Some other interesting announcements of the day:
- Live migration of a running server to GCP
- General availability of SQL Server enterprise
- Beta of .net available in App engine and container engine
- Beta of managed PostgreSQL
The major emphasis of day 3 was Google’s commitment to open source. It was really cool to see on stage Vint Cerf. Vint has contributed so much to the open standards of the internet that have been heavily adopted and used today.
He talked about his current work into the Interplanetary internet with NASA. He gave an overview of some of his achievements, and how much of an effort it was to get the TCP/IP protocol designed and open-sourced. Finally, they wrapped up with by announcing a new free tier model for GCP, in the picture above.
“This is open in every sense of the word” – Vint Cerf
On day 3 my highlight was the TensorFlow talk. Presented by Martin Görner, he explained some strategies for solving some different types of problems using machine learning on Tensorflow.
The most impressive thing was the demo he did showing what could be done when you train a model to create a new Shakespeare play. It was really interesting to see how the play evolved throughout each iteration of the model. It began by coming up with some jumbled text that didn’t really resemble English. By the end, we had a play that didn’t make much sense, but it was making up character names and constructing sentences which obviously fit in with the style of Shakespeare’s writing. It would be interesting to see what could be done with more training of the model or better algorithms for machine learning.
Martin also gave a demo training a model to write Python code. It started by looking nothing like Python code, but it eventually constructed some code that was obviously Python (minus a lot of syntax errors). It even recited the Apache software licence in comments at the top and wrote comments about what the code was doing although it made no sense.
Cloud Next ’17 was the first major conference that I have attended and I didn’t really know what to expect from it. I would have to say that, apart from a poor opening keynote, overall there were a lot of great presentations. Most of the sessions I attended were well thought out and really informative.
I have learnt a lot over the course of last week and I am even more eager to learn more about the many products and services available on GCP. I would have liked to see more major announcements, but the great breakout sessions made up for it. It’s quite surprising that Google didn’t delay the release of Cloud Spanner until the day 1 keynote. It could have made the day 1 keynote a lot more impressive. To sum up, overall it was a good conference. I can see that GCP will be a major rival to the likes of Amazon in the future, however, it is not quite there yet. I am looking forward to seeing what comes Next from GCP.