TWiGCP — “TPUs meet GKE, Stackdriver workspaces, and some lifehacking”

Here are your GCP news stories for this past week.

From the “Kubernetes news from Google and from the community” department:

• Cloud TPUs in Kubernetes Engine powering Minigo are now available in beta (Google blog). A single cluster creation flag, a YAML resource constraint one-liner, and TPUs meet GKE!
• Automating Canary Analysis on Google Kubernetes Engine with Spinnaker (Google documentation)
• KubeRBS for automatic Kubernetes rollbacks so you can sleep better at night! (blog.doit-intl.com)
• Kubernetes: ConfigMaps and Secrets (With Firebase) (medium.com)
• Skaffold 0.14 (github.com). New `skaffold init` feature

From the “Firebase offline and leveraging new App Engine runtimes” department:

• Multi-Tab Offline Support in Cloud Firestore! (Google blog). You can now turn on this experimental multi-tab offline synch with the Cloud Firestore Web SDK 5.5.0 release for Chrome, Safari and Firefox.
• Developing an App Engine service with Python and Cloud Firestore (medium.com)

Using Stackdriver Workspaces to help manage your hybrid and multicloud environment (Google blog). Organize workspaces by team function, by organization, or by environment.

From the “trust through transparency” department:

• Access Transparency logs now generally available for six GCP services (Google blog)
• Deleting your data in Google Cloud Platform (Google blog)

From the “if you needed reasons to use Google’s BigQuery, Dataproc, and Dataflow” department:

• Ethereum in BigQuery: how we built this dataset (Google blog). A follow-up to a popular blog post and how a BigQuery feature allows disassembling all Ethereum smart contracts in seconds.
• A flexible way to deploy Apache Hive on Cloud Dataproc (Google blog). An ephemeral cluster of preemptible VMs often means a cheaper solution too.
• How Distributed Shuffle improves scalability and performance in Cloud Dataflow pipelines (Google blog). Keep state storage and compute separate and connect them with a distributed shuffle layer.
• Moving to BigQuery (blog.hike.in)

From the “AI/ML tools, challenges, and GCP deep learning images” department:

• Introducing the Unrestricted Adversarial Examples Challenge (Google blog)
• The What-If Tool: Code-Free Probing of Machine Learning Models (Google blog)
• GCP deep learning images with preconfigured Tensorflow, PyTorch, Jupyter, Cuda and CuDNN (twitter.com)

From the “Lifehacker” department:

• The Google Cloud Developer Cheat Sheet (medium.com)
• Get your kid ETA from school using Google Home, Cloud Functions, Datastore, Maps Directions API… (blog.doit-intl.com)

From the “Beta, GA, or what?” department :

• [GA] Cloud SDK 216.0.0
• [GA] VPC internal DNS service
• [GA] Dataflow Shuffle (us-central1 and europe-west1)
• [Beta] Dataflow Streaming Engine (us-central1 and europe-west1)
• [Beta] Cloud TPUs in GKE
• [Beta] BigQuery Geospatial
• [Beta] Scheduling Queries

From the “all things multimedia” department :

• [Podcast] GCP Podcast episode #0146 is Google AI with Jeff Dean (gcppodcast.com)
• [Podcast] Kubernetes Podcast episode #020 is a conversation about Cloud Native Patterns for Ops, with Justin Garrison (kubernetespodcast.com)
• [Video] Serverless, Knative, Kubernetes, and You (acloud.guru)

This week’s picture is the updated (Google Cloud) Developer Cheat Sheet :

That is all for this week!
-Alexis