Even though Next 18 has come and gone, that doesn’t mean there aren’t new Google Cloud stories to get excited about! We have the biggest ones right here.
New Tesla P4 GPUs now available in beta
It’s well known that resource heavy graphical applications and machine learning applications gain a big boost from using GPUs for their work. About a year ago, Google Cloud introduced the option of using Nvidia GPUs for Google Compute resources with a single offering.
As of this week, there are now 4 different GPU options available in Compute Engine, at a variety of price points to match performance needs. The newest offering, just announced, is the Nvidia Tesla P4. It sits in the middle, offering a good balance of price and performance. It’s available in beta now, and is ready for anyone to use it (with the usual requisite beta disclaimers of course).
Istio goes 1.0!
At Google Next, Google talked at length about Istio, an open source framework that manages microservices across multiple platforms, both on-premises and cloud. It’s newest managed service, Cloud Service Platform, is itself build on top of Istio. As of last week, Istio reached the 1.0 milestone, which means that it’s out of beta and ready for production workloads.
Now, this wasn’t a Google-only effort. A number technology companies helped to develop Istio. Both Google and IBM were key co-founders of the project. Other tech companies such Lyft, Cisco, Red Hat, and VMWare have all contributed as well. Now that it is production-ready, I can’t wait to see how other organizations incorporate it into their hybrid application needs.
Repairing network hardware at scale with SRE principles
Finally, Google released a fascinating behind the scenes article of how their networking team manages and repairs their hardware architecture, using what’s known as site reliability engineering (SRE) principles.
In the article, we read how Google handles hardware failures and standardizes procedures to deal with their fleets of servers. The core principles of dealing with this scale of resources are to (1) treat servers as cattle instead of pets and (2) use automation whenever possible.
It is a fascinating dive into how Google maintains their infrastructure, and I definitely recommend checking it out.
Well, that’s going to wrap up this edition of Google Cloud weekly, I hope you enjoyed it, and we’ll see you next week!