GCP Updates | October 16, 2019

Posted on Oct 19, 2019


Cloud Dataproc – GPU support: GA

Boost the performance of workloads, including machine learning and data processing, by attaching graphics processing units (GPUs) to master and worker Compute Engine nodes in a Cloud Dataproc cluster. Choose from options including NVIDIA® Tesla® P100, V100, and P4 GPUs. Documentation

BigQuery ML – k-means clustering support: GA

Use unsupervised learning to aggregate your data into clusters and identify natural groupings. The k-means algorithm has been used for everything from understanding customer segmentation to computer vision and astronomy. Documentation


Dialogflow – agent validation: beta

Automatically check your agent for errors whenever you train it. Access the results from the Dialogflow Console or API, and choose to ignore or correct the errors to manage the quality and performance of your agent. Documentation

Dialogflow – regular expressions (regexp) for entities: GA

Now you can provide regular expressions to create agents that match data patterns – such as national identification numbers or license plates – rather than specific terms. Documentation

Dialogflow – fuzzy matching: GA

Spend less time and effort constructing entities – and simplify development. Entity matching requires an exact match for one of the entity entries. Fuzzy matching allows entities to identify matches of a value or synonym regardless of the word order. Documentation


Compute Engine – custom virtual network interface controller: beta

Enable higher network bandwidth with a more efficient delivery network for sending traffic to and from your VM instances. Specifically designed for Compute Engine, the gVNIC device and driver will work on all Google Cloud VMs except memory-optimized machine types. Documentation

Kubernetes Engine – container-native load balancing: GA

Enhance traffic visibility, performance, and scalability by creating services using network endpoint groups (NEGs). NEGs ensure requests to your service are distributed directly to the containers serving the requests. Documentation | Blog


Kubernetes Engine – usage metering: GA

View the usage profiles of Kubernetes Engine clusters and tie usage to individual teams, business units, apps, or environments within your organization based on namespaces and labels. Usage metering makes it easier to adopt Kubernetes in a multi-tenant environment. Documentation

Kubernetes Engine – containerd: GA

Use containerd instead of Docker as the container runtime for Kubernetes on Container-Optimized OS and Ubuntu to execute apps within containers and handle interactions with the underlying host operating system. Documentation



VPC Flow Logs – log sampling, aggregation, and metadata exclusion: GA

Balance your traffic visibility and storage costs – and lower monthly bills – by adjusting aggregation time interval, log-entry sampling, and metadata annotations to reduce the volume of data in your logs. Documentation


Stackdriver Logging – support for exporting logs to partitioned tables: beta

Improve BigQuery performance and gain greater control over your costs. Reduce the number of bytes read by a query by organizing the data you export to BigQuery into partitioned tables. Documentation


Kubernetes Engine – Binary Authorization: GA

Ensure that only trusted containers run on Kubernetes Engine with this deploy-time security control. You can require images to be signed by trusted authorities during the development process and then enforce signature validation when deploying. Documentation | Blog

Cloud Identity and Access Management – Policy Troubleshooter: beta

Use Policy Troubleshooter to examine all identity and access management policies that apply to a given resource, and determine whether a member has been granted permission to access that resource and, if so, by what policies. Documentation

便利そう ✨ 今度なんかに困ったときに使ってみたい。


Cloud Build – GitHub app triggers: beta

Configure automatic builds of code stored in GitHub to enable GitOps-like CI/CD workflows. This release includes enhanced security features, new API methods to manage triggers, and greater flexibility and support for a broader range of build configurations. Documentation