AWS re:Invent 2023 Recap

By Observe Team December 4, 2023

The Announcements

For many in our industry AWS re:Invent is the conference of the year. New services announced at re:Invent can have deep implications for businesses in a great range of sectors. The 2023 announcements were spread fairly evenly across categories compared to recent years where Compute and Machine Learning led the pack. Rather, this year the machine learning category was subsumed by Generative AI as AWS seems intent on catching up to some of its public cloud competitors on GenAI. A more comprehensive list of announcements from AWS can be found here, but here’s a few that caught our attention:

Generative AI

In September Amazon announced the general availability of its Bedrock service, which offers users access to a selection of tools for building their own Generative AI applications. They’ve continued to add to that service while also unveiling new GenAI services like their Q AI assistant.

  • Q, Amazon’s new AI chatbot. Q uses a variety of connectors to tap into data from all over an organization, but it’s also pre-trained on AWS data to better understand the nuances of the provider’s services. Q can help answer a user’s questions about their own business, or help them build on AWS services by answering questions specific to information in their documentation and the Well-Architected Framework.
  • Bedrock gets guardrails and more models. Amazon’s Bedrock service lets users select from a variety of foundational models, which now includes Llama 2, Stable Diffusion XL 1.0, and Anthropic Claude 2.1. Users can now also set guardrails to fine tune what models can’t or shouldn’t do.

Observability

Though well represented in the exhibition hall, Observability was not the star of the show in terms of Amazon’s announcements. However, there were notable additions to their native monitoring service CloudWatch including:

  • A new log class for CloudWatch. The new Infrequent Access log class is designed to make CloudWatch logs more cost-effective, albeit with the tradeoff of lacking some features of the Standard log class like Live Tail or data protection.
  • The preview of natural language queries for logs and metrics. CloudWatch isn’t known for having the most friendly UX. To make things a bit easier, users can now use natural language to generate queries for Logs and Metrics Insights. We’re a big fan of this kind of application for generative AI (in case our OPAL Co-Pilot might have already tipped you off).
  • Automated pattern analytics for CloudWatch Logs. CloudWatch now supports automatic recognition of cluster patterns among logs. Users can also compare query results from a select time interval to those of another time. Additionally, it can automatically flag anomalies in ingested logs.
  • The ability to to unify on-premises and multicloud metrics. This feature enables you to set up connectors that can pull metrics on demand from sources such as Prometheus or Azure Monitor. This means users can take metrics from sources on-premises, in AWS or in another cloud provider and unify them in CloudWatch if they so chose.

The Audience

Few events can rival the turnout that re:Invent commands. For context re:Invent 2023 hosted more than 65,000 attendees, while RSA this year had 40,000. re:Invent hosts thousands of sessions over the course of a week that cover the gamut of services in the AWS portfolio. It’s this wide range of topics that make for such a diverse audience. VPs, SREs and software engineers were common titles coming by our booth, and we had more visitors than ever before!

Observe Reinvent Booth
A core topic of conversation with attendees was the intersection of Observability and Generative AI, not surprising given the AWS announcements. The practicality of GenAI features (like O11y GPT Help and Slack Assistant) got a warm reception, especially for the ability to help users in various roles get up to speed with observability tooling quickly. Attendees were also interested in how features worked and where GenAI models were being hosted (for example the models behind OPAL Copilot and O11y GPT Extract are self-hosted by Observe). The topic of observability costs and frustration with legacy vendors was another common catalyst for conversation, as it is at every event we go to.

The Venue

The venue, as in past years, was the Venetian in Las Vegas. Familiar ground to many, but do indoor gondola rides ever lose their luster?

TLDR

Generative AI and LLMs are the current installment in the long running series of much hyped technologies under the AI umbrella. However, the wide ranging practical uses for GenAI have caught the attention of the industry. The updates we saw to its CloudWatch monitoring service reaffirms the commitment we at Observe made earlier this year to bringing GenerativeAI to bear on observability with features such as O11y GPT Extract, O11y GPT Help and OPAL Co-Pilot. And we have more AI-powered features on the way so stay tuned! The announcements from AWS will help to further cement the use of Generative AI and LLMs with a broader mainstream audience.

Undoubtedly, AWS re:Invent continues to be a world class tech event and testament to the pervasiveness of AWS adoption in the industry.

Want observability that can span all your cloud providers and offer all the advantages of Generative AI? Get started by signing up for the free trial of Observe here.