Google Cloud Platform Technology Nuggets — June 16–30, 2024 Edition

Welcome to the June 16–30, 2024 edition of Google Cloud Technology Nuggets.

Please feel free to give feedback on this issue and share the subscription form with your peers.

Containers and Kubernetes

In GKE, the Cluster Autoscaler (CA) is the component responsible for automatically adjusting the size of your node pools based on demand. Not that you have to deal with it directly but the team behind the component works on continuously looking at improvements to make it better. Some of the recent improvements include improved memory usage, less CPU waste, fast homogenous scale up and more. Check out the post for details. Some of the becnhmarking results and improvements are highlighted below:

Identity and Security

The first Cloud CISO perspectives for June 2024, three of the more promising use cases for AI in cybersecurity are discussed:

• AI for malware analysis
• Using AI to boost SecOps teams
• Using AI to scale security solutions

Google Cloud has launched of Cloud KMS Autokey in preview. Cloud KMS Autokey automates key control operations for CMEK while at the same time, incorporating best practices to reduce toil while managing your own encryption key. Check out a detailed article that provides information on how these features work in detail and how you can get started with it.

API Management

If Dialogflow CX is your choice of chatbot development platform, it is often recommended not to expose your bots directly and instead opt for an API Proxy layer that can manage and route your calls to your Dialogflow CX Agent. Apigee is a good solution that can help you do that. Check out the blog post on how you can configure a Reverse Proxy in Apigee to secure your Dialogflow CX Agents today.

Continuing with Apigee, Custom Reports in Apigee, allow you to get insights into your API. Check out this blog post that highlights several custom reports that you can consider to troubleshoot production deployments, optimize performance, and enhance your overall API strategy.

Machine Learning

There was a mega announcement around Vertex AI with significant advancements in models and enterprise platform capabilities with Vertex AI. Key announcements included:

• Gemini 1.5 Flash is now in GA with 1 million-token context window and extremely competitive pricing and performance.
• Gemini 1.5 Pro with 2 million-token context capabilities.
• Imagen3, Google’s latest image generation model is now available in preview for Vertex AI customers with multi-language support, built-in safety features like Google SynthID digital watermarking, and support for multiple aspect ratios.
• Claude 3.5 Sonnet model is available on Google Cloud.
• Context Caching feature is now available for both Gemini 1.5 Flash and 1.5 Pro.
• More factual responses via improved and expanded grounding facilities. More on that in the next section.
• Data Residency

Check all the details here.

Enterprises have clearly marked out grounding of the models as a key requirement to ensure that the experience that their customers/users see contains results that are factual. Towards this, there are key improvements/expansions been down to the grounding capabilities in Vertex AI. These include:

• Grounding with Google Search feature will introduce dynamic retrieval, that will intelligently select when to use Google Search results and when to use the model’s training data.
• Vector Search will offer hybrid search, now available in Preview.
• Upcoming Partnerships with data companies to provide further grounding in terms of factual data.
• A high-fidelity mode that is designed to further reduce hallucinations.

Check out the blog post for details.

Anthropic’s newly released model, Claude 3.5 Sonnet is making waves across the Generative AI world with its capabilities and new benchmarks for reasoning, undergraduate-level knowledge, math, and coding. The model is available on Vertex AI via the Model Garden repository. Check out the blog post on its capabilities,which customers are using it and how you can use it today via Vertex AI.

Gartner® has named Google as a Leader in the Magic Quadrant™ for Data Science and Machine Learning Platforms. Check out the blog post and download the complimentary 2024 Gartner Magic Quadrant™ for Data Science and Machine Learning Platforms.

Databases

If you’d like to get a summary on various services in Google Cloud Databases in the month of June, you can check out this great summary blog post titled “What’s new with Google Cloud databases — June 2024 edition”. If you are working with Google Cloud Databases, this is a must read.

There are several new features for Cloud SQL for MySQL, available in Preview. These include:

• Support for storage and similarity search of vector embeddings. K-nearest-neighbor (KNN) and approximate-nearest-neighbor (ANN) search between embeddings is supported.
• Gemini in Cloud Databases is available throughout now. You can use Index Advisor, address performance issues with Active Queries and more.

Check out the blog post for details.

Log-based Metrics is a neat feature available in Cloud Logging, where you can observe your logs, identify patterns in the log message and turn them into metrics. You can then use these metrics to configure alerts that can get triggered based on certain threshold values that you would like to define. The logs could be from various Google Cloud Platform services and specifically it would be useful to create log-based metrics based for your PostgreSQL and AlloyDB services. Check out this blog post that provides an example of how you can do that. If you are into monitoring for certain exceptions vis-a-vis usage of your database services, this would be a must-add feature.

Networking

With the rise of Generative AI applications, organizations are looking at ways to address how best to host and serve these workloads to optimize of costs and cloud resources like CPU, memory, etc. If you are looking to assume that these are still web applications that are front-ending a model inferencing endpoint, that might not be accurate. Here is a good article that starts off with highlighting the differences between traditional web traffic and AI traffic and then talks about various innovations available across Google Cloud Networking and baked into Vertex AI, that you can look at today.

Google Cloud’s IPv6 Hybrid Connectivity portfolio has seen some key announcements:

• IPv6 BGP sessions
• Partner Interconnect IPv6
• IPv6-only HA-VPN

Check out the blog post for more details.

Struggling with having your services up and available behind Google Cloud Load Balancing? How about some some troubleshooting tips for that? The article takes you into understanding health checks, the tools available at your disposal and tips. Check it out.

Moving on from troubleshooting to benchmarking. In what is likely to be a series of whitepapers on best practices for network performance and benchmarking, the first article cover best practices for benchmarking Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) bulk flows. Check out the blog post for a series of white papers on this topic.

Data Analytics

Vector Search in BigQuery enables semantic similarity search in BigQuery. A fantastic use case that gets enabled via this feature in BigQuery is to analyze logs and asset metadata stored in BigQuery tables. So if you are funneling your logs into BigQuery, you are all set to analyse a log entry, identify if it is an anomaly, compare it across previous instances and more.

Check out this blog post that does a great job at highlighting differences between a text-based and semantic search, understanding the Vector Search feature in BigQuery and then looking at real-world cases in how you can use them.

Here is an interesting use case. What if you had to finding and understanding reviews in a customer’s preferred language across multiple languages. Continuing with the previous news item of having Vector Search supported in BigQuery, the solution involves usage of BigQuery multilingual embeddings, vector index and vector search, to let customers search for products or business reviews in their preferred language and receive results in that same language. Check it out.

Gartner® has named Google a Leader in the 2024 Magic Quadrant™ for Analytics and Business Intelligence. Check out the blog post and download the complimentary 2024 Gartner Magic Quadrant for Analytics and Business Intelligence Platforms.

Cost Management

All cloud costs should be visible and allocated, spend should be efficient with no waste, and there are of course no surprise costs. That has been the philosophy of Google Cloud FinOps. At FinOps X 2024, several interesting announcements. Some of them included an update to the BigQuery view to match the latest FOCUS v1.0 Specification GA release, integrates carbon footprint reporting and Scenario Modeling for CUDs. Check out the blog post for more details.

The previous paragraph talked about FOCUS. What is FOCUS? It stands for FinOps Open Cost & Usage and is a technical specification that normalizes cost and usage billing data across cloud vendors. Think of it as a unifying specification for cloud billing data. Google Cloud has introduced:

• A new Looker Template for Focus v1.0 GA
• An updated BigQuery view for Focus v1.0 GA

Check out the blog post for more details.

Developers and Practitioners

Google Cloud supports six deployment archetypes: zonal, regional, multi-regional, global, hybrid, and multicloud. What is a deployment archetype? It is an architectural model that you use as the foundation on which to build cloud topologies that meet the business and technical requirements of your application. It influences your availability, latency, performance, and cost requirements.

Check out this blog post that dives deeper and provides reference architecture material for each of them.

DevOps and SRE

If you are into Operations, you would be grappling with the task of understanding what Generative AI is and how it could apply to your daily tasks with Google Cloud or any other environment. Here is a great list of resources curated for a SRE role, to come up to speed with fundamentals of Generative AI and several codelabs that help you experience the power of Gemini. Take a look.

Learn Google Cloud

Have you joined the Cloud Innovators program? The program provides you 35 credits every month to use towards courses and hands-on labs.

Join the Innovators program at no cost today!

Stay in Touch

Have questions, comments, or other feedback on this newsletter? Please send Feedback.

If any of your peers are interested in receiving this newsletter, send them the Subscribe link.

Want to keep tabs on new Google Cloud product announcements? We have a handy page that you should bookmark → What’s new with Google Cloud.