[San Francisco, United States = DigitalToday reporter Chi-gyu Hwang] Snowflake acquired Observe in January and moved fully into the observability market this year. That puts it in competition with Cisco-owned Splunk, Datadog and New Relic.

Snowflake acquires observability startup Observe: "worth $1 billion"

Jeremy Burton (제레미 버튼), the founder of Observe and now general manager of Snowflake’s observability business, said in an interview at Snowflake Summit 26 in San Francisco from June 1 to 4 that Snowflake’s data platform would be a meaningful foundation for strengthening a competitive edge in the observability market.

"Observability is inherently a data problem, and it was difficult and expensive because data was fragmented," he said. "Through Snowflake, we can solve this problem," he added.

He said the existing observability market has a structural problem. Splunk started with log analysis and added monitoring and application performance management through acquisitions, while Datadog grew around time-series monitoring. New Relic expanded from application performance management. Because they started from different architectures and then added functions, he said, the structure forces users to use 2 or 3 products together.

"If you were building everything as one system from the beginning, you would naturally have chosen to put the data in one database," he said. "Snowflake can take this approach because it is also a relational database while processing semi-structured JSON, unstructured data and time-series data," he added.

He also cited cost competitiveness as a strength. "In a market where Splunk charged $5 per gigabyte per month, the idea that we could lower it to 50 cents by using Snowflake’s architecture that separates storage and computing was the starting point for founding Observe in 2018," Burton said. "To make the observability cost structure sustainable, you ultimately need to own the data platform directly. A model where you pay for an external data platform does not make economic sense," he added.

Burton also stressed the use of AI agents. "If you do not give AI agents context, they wander and burn through the token budget, but if you provide well-organised context, they can reach accurate answers at a manageable cost," he said. "The data structuring work Observe has built up over a long time can deliver greater value to AI agents than to people," he added.

At this Snowflake Summit, Snowflake officially launched PG Lake and also announced a public preview of Data Mirroring.

PG Lake is a set of tools for moving data from Postgres to a data lake or to Apache Iceberg, and it is also set to be released as open source. Data Mirroring automatically replicates data in a Postgres database to Snowflake with the push of a button. Burton said, "In Ericsson’s case, we cut data movement from 2 days to 7 minutes."

On future strategy, Burton pointed to integrating and analysing observability data and business data on the same platform.

"When a service outage lasts 2 hours, you should be able to ask immediately what the business impact is, without separate analysis," Burton said. "If telemetry data and business data are in the same schema, you can answer this question right away," he added. He said he expects the development of Apache Iceberg-based open formats and open catalogs to make such integration much easier within 1 to 2 years.