Snowflake Inc. today announced that it’s buying Observe AI Inc., a well-funded startup with an eponymous observability platform.

The terms of the deal were not disclosed. When rumors of the acquisition first emerged last month, The Information cited sources as saying that it values Observe at about $1 billion. That’s about two-and-a-half times the amount of funding the nine-year-old company raised from investors.

Observe’s platform helps developers find the root cause of application slowdowns, website outages and other technical issues. It can also collect data on the more routine aspects of a company’s technology operations. For example, a software team could use Observe to track the inference costs incurred by a large language model.

Snowflake plans to integrate the company’s observability software into its namesake cloud data platform. The effort is set to particular emphasis on an Observe component called AI SRE, for site reliability engineer.

As the name suggests, AI SRE is an artificial intelligence chatbot that helps developers diagnose technical issues. Users can not only ask the AI to uncover the cause of a malfunction but also customize how it goes about the task. For example, AI SRE could be instructed to focus solely on error logs from the past 24 hours when analyzing a newly detected outage. 

AI SRE is powered by a data management engine called the O11y Context Graph. According to Observe, the engine ingests hundreds of terabytes of telemetry per day from customer environments. It links together related pieces of data and creates indexes, collections of shortcuts that speed up information searches. The O11y Context Graph also generates materialized views, cached copies of query results that cut loading times.

Observe’s platform keeps the data it processes in Snowflake. According to the companies, the acquisition will ease troubleshooting tasks for customers by enabling them to implement longer retention windows.

A retention window is the amount of time that passes before an enterprise deletes its archived telemetry. Historically, organizations had to delete telemetry fairly often to avoid excessive storage costs. That practice can complicate troubleshooting efforts. If a technical issue first emerged two months ago but the affected company only keeps error logs for only one month, engineers may struggle to find the root cause.

Snowflake compresses data and keeps it in low-cost object storage services to lower customers’ infrastructure bills. According to the company, its platform’s object storage features will enable customers to retain telemetry for longer than would otherwise be practical. That means more information will be available for incident investigations. 

“An organization’s ability to remain resilient is strictly limited by how much data it can afford to ingest, how it navigates the myriad of formats and silos, and the speed at which it can reason across it,” Christian Kleinerman, Snowflake’s executive vice president of product, wrote in a blog post today. “Snowflake and Observe will empower customers to manage enterprise-wide observability across petabytes of telemetry with a modern, scalable architecture, enabling them to run production applications and agents with greater confidence, without sacrificing data to control costs.”

The deal comes about two months after Snowflake’s previous acquisition. In mid-November, it inked a deal to buy Datometry Inc., which developed a tool for moving data to Snowflake environments from rival platforms. The startup also created an open-source relational database called OpenDB.

Photo: Snowflake