Google Cloud Next '19 in San Francisco has been my second Google Cloud Next event, the first one being last year's London edition. While back in London I could only see the last day, this time I've had a pleasure of attending full three days of conference (Thanks, CodiLime!). Lots of announcements, lot's of on- and off-site events, lots of interesting sessions. In fact, there were many times when I simply couldn't decide where to go and what to see, but it's all good once you overcome your FOMO since the sessions are also available on the GCP YouTube channel. Here are some of my personal thoughts about the event.

All aboard the Next train!

Highlight 1: Everything Anthos

The big thing of the first keynote presentation. Google Kubernetes Engine (GKE) on-premises. Multi-cloud GKE.

Anthos (formerly Cloud Services Platform) lets you use Google's Kubernetes managed solution anywhere you want. It seems like a perfect solution for situations where the need for hybrid environment together with difficulties of managing your own Kubernetes cluster were stopping you from embracing the cloud. Together with GKE come all the other cloud services, so you can use products like Stackdriver, Cloud Build freely while having control over your infrastructure. Neat, huh?

And you know what? It works on AWS and Azure, too. (You can see how GKE compares to other native managed Kubernetes services and why you would want to use it here.)

It's a big thing, but surprisingly there wasn't as much coverage of it (or did I miss it?). I guess we'll see more in the months to come.

Highlight 2: Gitlab migration from Azure to GCP

I believe there isn't too many people who don't know what Gitlab is, but in case introduction is needed - Gitlab is a git repository hosting with a lot of additional features, like built-in CI/CD, docker registries, security checks, license compliance management, wikis... There's too much to say here about what it is and what isn't so I encourage you to have a look at their extensive documentation.

Other than being a service provider, Gitlab is a champion of being open in so many different ways. First of all, Gitlab as a platform is following the open core model, where the premium version is an extension of the FOSS edition. What is important, the FOSS Gitlab is not stripped of any essential functionalities for majority of users, so it's not forcing anyone to use the Premium edition just to get work done. Moreover, they openly own and show how they solved any issues that are happening while running this (pretty big!) business (such as the infamous database incident).

A session that was a real highlight of the event took place on Wednesday and covered the topic of migrating the SaaS instance of Gitlab - gitlab.com - from Azure to GCP. I highly recommend watching this session on Youtube: link.

For me, it was the ultimate example of dogfooding. For those who don't know what dogfooding means: it's using the software that you've created on your own, to make sure that it's working the way it should. In case of Gitlab, they created Geo: a tool which allows for creating Read-only mirrors of your GitLab instances. Then, they used it (along with other things) to do a full migration of their infrastructure. I don't want to spoil this awesome presentation for you, so just go and watch it, seriously. If you're more of a reader, have a look at Gitlab's migration doc here.

Just don't skip it.

Highlight 3: Cloud costs and efficiency track

New tech is cool and all, but if it fails to bring value to your product (and revenue for the stakeholders), it is not going to be used. One way to improve that, very often overlooked when trigger-happily firing new stuff, is through optimizing the costs. The topic has already been covered in little bits on the GCP blog, but multiple sessions were showing what's new and hinting at how to be smart about your money in the cloud. Be sure to explore the cost-management documentation as well as the following sessions:

• Establishing Financial Governance Controls on GCP was a session showing how to control your spending using budgets, quotas and smart notifications connected with Cloud Functions to do your accounting and governance automatically. I feel that this was an extension and a demo of a Cloud Blog post from last year. Additional information: Budgeting API is coming soon to Alpha and service usage/quota API is already in available.
• Beyond Your Bill: Getting More Details on Cost explained some more advanced patterns that you can use beside the basic export of detailed data to BigQuery. Tracking Cloud Datastore usage by following API endpoints? Check. Tracking your spending on project and client-level? Check. Cool visualizations in Data Studio? Check. Spend lovely 34 minutes watching this session? You betcha.
• Automating Cost optimizations on GCP was kind of a spiritual successor of the first session mentioned here. Once we identify the issues that trouble our infrastructure, it's great to follow the examples shown in this session to automate your environment cleanups and analysis. Check out the mind exercise at 7:00 - I bet you've thought "There's no way someone else has had this issue" and then you hear your own issue from the crowd... We've all been there. Don't reinvent the wheel, get inspired.
• GKE Usage Metering: Whose Line Item is This Anyway was a presentation about GKE's new feature (currently in Beta) described on the blog as well - usage metering. It allows you to track spendings for clusters per-namespace or per-labels. Super useful for multi-tenancy use cases, only a bit less so for specialized clusters. The session was based on Spotify's use case as they're moving to GKE during Helios-to-Kubernetes migration.

That's a wrap!

It took me a few days to finish this wrap-up, but I hope it's not stale yet :-).

Google Cloud Next 19 was a fantastic event and I've had a great pleasure to be there. I've certainly learned a lot of new things and have had a chance to meet people from lots of different companies, which are all connected by at least one thing - being cool with the cloud. I highly recommend joining Cloud Parties like this one whenever you've got the chance.

Have you been there? Do you want to share your thoughts on any of the sessions? Feel free to tweet me or contact in any other way.