Over 30,000 people flocked to Las Vegas to see the latest and greatest from Google Cloud and its partners at Google Cloud Next 2024. As a long-time Google Cloud partner and recipient of two Google Cloud Technology Partner of the Year awards this year, we were there in full force to showcase our unified observability and security solutions and engage with the Google Cloud community. From keynotes to breakout sessions to the expo floor (our team was partial to the Innovator’s Hive), there was no shortage of things to see. Here are some of the key themes and announcements from this year’s event.
Google’s primary focus at this year’s event was the improvements they have made across their AI stack, particularly their foundation models. Gemini 1.5 Pro entered public preview in Vertex AI, becoming Google’s most powerful model yet, with multimodal input support and a context window of up to one million tokens (which will be expanding to two million later this year). And with new variants of their open source Gemma model and the general availability of Imagen 2.0, Google continues to provide ways for its users to capitalize on the power of AI.
In an effort to make Gemini even more powerful for Datadog users, we have collaborated with Google to train Gemini on our public documentation, enabling it to assist users with Datadog setup and usage (as outlined by Datadog’s Sri Raman and Jason Hand and Google Cloud’s Prithpal Bhogill and Merlin Yamssi in a joint breakout session).
These foundation models are made available to Google Cloud users in Vertex AI’s Model Garden. And for users looking for a lower-code way to put these foundation models to use, Google introduced their new Vertex AI Agent Builder, where users can build AI agents via conversational prompts that improve customer experiences, increase employee efficiency, and more.
At Datadog, we’re helping our users reliably invest in AI by giving them insight into the health and performance of their LLMs. Our Vertex AI integration provides users with inference and infrastructure metrics for both out-of-the-box models and custom models trained on their data. This enables side-by-side model comparison and allows users to ensure their models meet their performance standards before deploying them into production.
Complementing these improvements to Google’s AI models were new investments into their AI Hypercomputer. Google’s most powerful TPU yet, v5p, entered general availability. Their new A3 Mega VMs, which offer double the network bandwidth of A3 VMs, are soon to follow. And with the releases of Arm-based Axion GPUs (for which Datadog is a committed partner) and NVIDIA’s Grace Blackwell GPUs on the horizon, Google Cloud users will continue to get the most performant infrastructure to train their models.
As customers start taking advantage of this new hardware, scaling their cloud usage and spend, they need tools to ensure they’re getting the most value for their spend. Observability is critical to an effective cost management strategy, and with new Google Cloud support, Datadog’s Cloud Cost Management gives organizations granular visibility into their Google Cloud spend. As Datadog’s Scott Mabe outlined in a breakout session at Next, this visibility provides developers with greater context into their cost drivers and empowers them to take effective action to reduce it.
Given the growing number of cybersecurity attacks and the significant consequences of a breach, securing cloud environments is an increasingly important challenge for businesses. At Next, Google highlighted a number of ways its investments in AI can be used to secure your Google Cloud environment. The Vertex AI Agent Builder can be used to build cybersecurity specific agents that detect, prevent, and help resolve security threats. For more out-of-the-box functionality, Google has embedded Gemini into products like Google Threat Intelligence, Security Command Center, and Workspace to dig into cybersecurity threats, evaluate cybersecurity posture, and continuously identify and protect sensitive data.
At Datadog, we believe observability and security go hand in hand. In his breakout session, Rory McCune discussed how a unified DevSecOps mindset helps developers catch security issues in their code deployments and gives security engineers greater context during remediation. With integrations with key Google Cloud services like Google Cloud Security Center and Google Cloud Armor, Datadog can aggregate observability and security information from across your Google Cloud environment, centralizing your security monitoring and enabling your teams to more effectively resolve security issues.
It’s become clear that one of the strongest use cases of AI is improving engineer satisfaction and efficiency. To that end, Google introduced two code assistance tools at Next. Gemini Code Assist helps developers complete their code and generate new functions on demand with awareness of their full codebase. Google also introduced a new variant of their open source Gemma model—CodeGemma—which will provide developers with powerful, yet lightweight code generation and completion capabilities.
Another key component to improving the developer experience is streamlining the experience of releasing finished code. In Ajuna Kyaruzi’s session at Next, she shared how Datadog makes heavy use of our own CI tools—including CI Pipeline Visibility, Test Visibility, and Intelligent Test Runner—to address pipeline performance issues, improve developer efficiency, and reduce flaky tests.
This was just a subset of Google’s announcements and the ways that Datadog can monitor and secure your Google Cloud environment. To learn more about how Datadog can help you get the most out of your Google Cloud investments, refer to our solutions page. You can see our documentation to get started. Or, if you’re not already a customer, sign up for a 14-day free trial. We’re looking forward to seeing you all at next year’s event!
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。