Google Cloud Technology Nuggets — August 1–15, 2022 Edition

Welcome to the August 1–15, 2022 edition of Google Cloud Technology Nuggets.

Consolidated reference docs at cloud.google.com/sdk

If you have been scratching your head to find consolidated page for client libraries and tools, these have now been put under a single page: cloud.google.com/sdk. You will find the tools and specific programming language libraries all under this one URL. In addition to that, there are several other features that help you navigate this information easily: filters, search, automated document generation with each release, direct code edits and more. Check out the blog post for more details and bookmark this page.

Infrastructure

In a move that shows continued investment to address the growing Cloud market in Asia Pacific, 3 more Google Cloud regions are planned: Malaysia, Thailand, and New Zealand. Check out the blog post for more details. Check out the Global Infrastructure page for information on various regions and services.

Customers

A Clinical Decision Support System (CDSS) is complex to build. It requires an ability to locate and extract the medical entities that are present in the clinical notes, medical journals, discharge summaries, etc. In an interesting customer case study, Google Cloud worked with Apollo 24|7, one of the largest multi-channel digital healthcare platform in India, to build the key blocks of their CDSS solution.

As the blog post states, the solution was designed to “parse the discharge summaries and prescriptions to extract the medical entities. These entities can then be used to build a recommendation engine that would help doctors with the “Next Best Action” recommendation for medicines, lab tests, etc.”

Check out the case study that highlights the datasets used, the models evaluated, performance metrics and more.

Containers and GKE

If you have been on the journey to adopt Kubernetes as a platform for your application, what are the key traits of this platform? As per the blog post, Automation and Scale are the two key traits of a Kubernetes platform and with the innovations done by Google Kubernetes Engine, both of them are within reach.

When it comes to Automation, a fully-managed Autopilot mode of operation gives you a secured, production ready cluster in minutes. There are key other features like Day 2 Operations, Cost Optimization strategies, service mesh and more that ease in managing the environment.

On the scale characteristic, GKE can support 15,000 node clusters.

Security

Google Cloud Storage is a key service that integrates with multiple Google cloud services and is often the first step to getting data into Google Cloud. In a key development and keeping in line with data security and privacy requirements, it is now possible to de-identify data in Google Cloud Storage by configuring a Job that can scan your data and then perform one of many actions: redaction, replacement, masking, etc. Check out the blog post for more details.

Did you know about 5 key security and compliance features of Cloud Logging that can be used during a security audit? Check this out.

Continuing on Security, think about centralizing and managing access to different lines of business applications running across teams, departments, environments and more. This is exactly what Corp Engineering in Google, which manages internal applications needed from a solution. They went with Anthos Service Mesh.

Check out this interesting study of ASM features and how it fit the bill.

Databases, Analytics and Machine Learning

There are a couple of posts around BigQuery that are worth reading:

BigQuery is an excellent tool and here’s an article that is an authors take on some interesting features that they found. Maybe you will pick up a thing or two, that you didn’t know BigQuery had?
BigQuery costs can easily spiral out of control. This post goes into great detail to highlight key areas that you can tap into to control costs. It also provides training courses at the end to learn more in the area.

Data Studio is now available as a Google Cloud Service, which means that it will be available under the same terms as a Google Cloud Service. This ensure that Google supports the same privacy and security commitments for Data Studio as for other Google Cloud products. Check out the post for more details and specific steps to enable this.

What knowledge and skillsets are needed to help an organization succeed in their Machine Learning initiatives. What does a ML Engineer do and the skills that they should possess? What are the courses that are currently offered to help train them? This post has the answers.

Are you looking to use Vertex AI Training with large datasets that you have present on shared file systems? You can now use NFS shares such as Filestore. Check out this post that gives step by step details on using Filestore and train a model with your custom training code.

Operations

There are times when you would want the Alert Policiesi n Cloud Monitoring to not fire or in other words go in Snooze mode. This can be useful if the development team would not like to get paged during non-business hours, avoid alerts to be fired during a planned maintenance or even during an outage. The Snooze feature available in Cloud Monitoring makes this possible. Check out the post for more details.

Cloud Logging has a powerful Query language that lets you filter logs specific to resources, environments and more. What would be really cool would be to have templates for query logs across different kinds of services. For e.g. how about a built in query that allows you to search for Kubernetes node container runtime logs? The Query Library is meant to exactly do that by providing you to multiple pre-built queries that you can just preview/customize/save to speed up your search for specific logs across resources. Check out this post for details.

Cloud Deploy has seen significant updates in a recent release:

Auto-generation of Skaffold configuration file. This when done early in the cycle can help both the development and operators familiarize themselves with Google Cloud Deploy.
Delivery pipeline suspension and abandonment features.
Release Inspector tool, which makes it easy to review application manifests and compare against releases and targets within a delivery pipeline.
VPC-SC support for Cloud Deploy
Support for additional Cloud Deploy regions

Check out the blog post for more details.

Developers and Practitioners

Functions as a Service does not need any introduction and Google Cloud Functions has been available to help enable the programmable cloud for a while now. As developers used this service, 4 key areas emerged that needed to be addressed: cold starts, latency, connecting disparate sources, and managing costs. Cloud Functions 2nd Gen, is now available in GA that as a response to those requirements. It provides powerful and efficient compute options, granular controls for faster rollbacks and new triggers from over 125 Google and third-party SaaS event sources using Eventarc.

Check out the blog post for more details and documentation to get started.

Developers like to build out features and release them as quickly as possible. How do you manage this pace while making sure that the security rules are in place? Security guardrails or preventive security controls is what is proposed in this blog post. Infrastructure as Code plays a key role over here in addition to native features that are available in Google Cloud ranging from Organization Policy, VPC-SC, Cloud IAM. Do check out the Security Foundations blueprint.

EventArc is fast becoming the glue to tap into various events across Google Cloud services and then orchestrating application flow on top of that. Not just native Google Cloud services, but 3rd party event sources are now supported in EventArc. The first set of integrations allows Google Cloud to tap into solutions from Tech Partners like Datadog, ForgeRock, CloudGuard and Lacework. As an example, you could be using Datadog to monitor the number of Compute Engine instances, if they fall below a threshold, an event is raised and delivered to a destination in Google Cloud via EventArc. The destination can then drive a downstream functionality powered by Cloud Functions, etc to perform a specific action i.e. launch more instances, etc.