Google Cloud Platform Technology Nuggets — Mar 1–15, 2022 Edition
Welcome to the March 1–15, 2022 edition of Google Cloud Technology Nuggets.
If you have been confused about the fact that the gcloud CLI was referred to as gCloud SDK, there has been a rebranding to share a more consistent messaging when dealing with command line tools (CLI) and SDKs.
- Renamed our set of command line developer tools (gcloud, bq, gsutil, kubectl, etc) to Google Cloud CLI to make clear that this is the command line interface for working with any of our products and services.
- Our Cloud Client Libraries are now part of Google Cloud SDKs. Languages supported are Java, Go, NodeJS, Python, C#, Ruby, and PHP. These client libraries are now recommended for all new projects.
- By structuring around language, each SDK now has language specific tools and frameworks too.
For more information, check out the blog post.
What does a modern data platform, on fully managed services, with diverse datasets and processes running on top of this to tailor data for various stakeholders, look like? Leading beauty products provider, L’Oreal had some clear requirements when it went about building their data platform. These requirements ranged from No-Ops, secure, Event Driven architecture and more. The platform required ingestion of data via APIs and other applications but needed to support 1000s of user flows that curate data for the users. The high level architecture diagram shown below is based on various serverless offerings like Eventarc, Cloud Run and Cloud Functions (2nd gen), Cloud Workflows and more. The result as quoted from the article — “100TB of production data in BigQuery and processes 20TB of data each month. We have more than 8000 governed datasets, and 2 millions of BigQuery tables coming from multiple data sources such as Salesforce, SAP, Microsoft, and Google Ads. “
Read more on this case study.
Containers and Kubernetes
Knative, an open source framework that provides building blocks to run serverless containers, is now a part of Cloud Native Computing Foundation (CNCF). Since its inception in 2018, Knative has been developed by Google in close conjunction with IBM, Red Hat, VMware, and SAP and is the most-widely installed serverless layer on Kubernetes. Check out the Knative documentation page to learn more and for those of you interested in Kubecon EU, there are plans to hold a Knative Day.
DevOps and SRE
Google Managed Prometheus is now available for everyone. If you are using self-managed Prometheus and are facing challenges scaling Prometheus, this managed service is the right one to consider. As the architecture indicates, it is based on the time-series database Monarch and one of the key things to note is that it plays well with existing Grafana and Prometheus Alert Manager and other tools.
One of the best ways to experience this would be the installation guide for setting up a managed collection in your existing GKE installation and see Google Managed Prometheus in action. The guide is available here.
Every organization wants to be a data-driven organization but current challenges around modernization coupled with existing technical depth has slowed down the pace as the blog post suggests. The post is a good summary of how Google Cloud provides databases that are both modern and are powered by best-in-class infrastructure, security, extensibility and connectivity to other Google Cloud services, Open Source compatibility and more. This is important to ensure that your architecture is one that will stand the test for several years. A good refresher article to get an understanding of the range of databases that are there on Google Cloud and specific use cases that are best suited for them. Check out the companion PDF titled “Make your database your secret advantage with Google Cloud”.
Part 2 of the Data Governance series is out. This part focuses on the specific tools available in Google Cloud to achieve the key areas around Data Governance, some of which include: Auditing, Data Access Controls, Discovery and more. It is recommended that you check out Part 1 of the series too.
A key challenge if not the biggest challenge in AI is to productionize AI Models. A couple of Google papers have written about this and an interesting blog post covers some of the best practices that you can employ in bringing these models to production. The post focuses on Vertex AI and except for model development, it goes into all aspects of the MLOps framework. This is an excellent resource for any ML practitioner. The material is made available on Github and the post guides you through the data that is there, along with training, evaluation and deployment of the model.
Digital assistants are now a key part of online businesses. However, it is important for your brand to retain its own unique identity and this is especially important when it comes to voice, where the various options available to you are synthetic voices. With the release of Custom Voices in Google Text-To-Speech (TTS) API, you can now use your custom voices as part of the API. You submit your audio recordings as per a guidance document and relevant process checks and then once the model is trained, you can reference this model id during the invocation process and the API will use the specific custom voice model during the Text to Speech synthesis.
Serverless App Development
If Mathematics is your area of interest, you would be keeping track of March 14, which is celebrated at Pi Day. Three years back, Emma Haruka Iwao broke the Guinness record for most calculated digits of Pi (31.4 trillion digits). To celebrate Pi Day this year, Emma and fellow Developer Advocate Sara Ford have written up a blog post that describes how they are attempting that challenge again but this time with a new architecture, based on Google Cloud serverless compute technologies.
It is a great example of how some of the changes in the 2nd generation of Cloud Functions (larger instances, longer running times) coupled with a novel algorithm, can push the limits of what serverless architectures can achieve. Hidden in the blog post is also an interesting migration of the previous API off GKE and the cost savings that resulted from it. Do read the blog post.
Let’s learn about GCP
The series on Anthos keeps getting better with every subsequent post. The latest installment in the series (Part 7) covers Anthos Marketplace applications. Take a look at the entire series and bookmark it for upcoming installments.
Stay in Touch!
Have questions, comments, or other feedback. Do send it across.
Looking to keep a tab on new Google Cloud product announcements? Check out this handy page that you should bookmark → What’s new with Google Cloud.