Snowflake buys Observe to make 'Days Since Last Outage' counters obsolete
Nice idea, because its own cloudy services keep wobbling
Analytics outfit Snowflake is buying telemetry data platform Observe to help its customers discover and mitigate IT issues before they cause downtime. It announced the deal on the same day its own services experienced a “major outage.”
Carl Perry, head of analytics at Snowflake, told The Register that the company plans to use “all tooling at its disposal” to mitigate its own downtime incidents,” but has not yet added Observe’s products to its repair kits.
“We haven’t been using them yet. We evaluated using them. We think they’re a great product. It actually increased our confidence when we looked at the acquisition, given our experience and our evaluation personally, of how it can be effective here at Snowflake,” he said. “That just gave us yet another positive sign that this is something that would be amazing. I think the main focus, though, is our customers.”
Snowflake did not disclose the purchase price nor establish a timeline to close the deal.
Perry said enterprises collect massive quantities of data, but the high cost of storage means some ditch telemetry that could help to predict and mitigate downtime.
“People lose visibility into critical data because they just can’t afford to store it,” Perry said. “What Observe does is it inverts that process. Like ‘Lets just use cloud storage, which is the least expensive storage mechanism on the plant, and it has almost infinite scale, and the cost is incredibly low for customers, and then we’ll build the experience on top of that, leveraging Snowflake as the core data platform and then adding in observability specific functionality.’”
Observe’s method has won it many admirers. In July 2025, the company announced a fresh funding round that brought in $156 million, some of it from Snowflake.
At that time, Observe said it had tripled its revenue and doubled its enterprise customer base in a year, and processed 150 petabytes of telemetry data. The company said its customers include large enterprises, SaaS platforms, and AI native companies, some of whom ditched Splunk, Datadog, and Elasticsearch and started using Observe instead.
“I think observability has been a big problem for customers for a long time,” Perry told us. “And actually, I think you see that more and more companies are leaning on needing observability as their application stack explodes and then AI gets plugged into it, and their AI agents, AI services, AI apps being built. I think it’s an even bigger need now than it was in the past.”
Talking the walk
Snowflake’s motive for buying Observe is to help customers reduce outages, but it had two of its own on Wednesday and both impacted customers.
One was a degraded performance issue that prevented customers from accessing Snowflake features while using Snowsight, the company’s web interface.
Additionally, customers in its Azure West region experienced a “major outage” which also left customers unable to access Snowflake services and Snowsight.
Each issue was resolved in under two hours, according to Snowflake’s incident history page. The company also had two incidents in December, and four in November.
Snowflake rival Databricks also had an outage yesterday and another pair in December 2025, but does not list earlier incidents. Databricks’ incident notification pages are also less detailed compared to Snowflake’s, and it asks users to submit a ticket to learn about the cause of disruptions.
Snowflake describes incidents on its status page.
”Running a service in the cloud and being as transparent as possible can be “painful at times,” Perry said.
“It’s painful in the moment, when we talk to customers about an incident for sure. Because again, they took downtime,” Perry told The Register. “They had problems where they couldn’t access their data, or do some specific activity or operation. And we really feel poorly in that moment, but making sure they understand why it happened, what we’re doing, and showing that we actually did the things we did to stop those types of incidents from occurring ends up actually driving better relationships and trust between us as a company and our customers.”
Perry’s career has taken him through Amazon as development manager for AWS’s S3 API and front end services, to Microsoft where he was the group product manager responsible for launching PowerBI, and finally head of analytics at Snowflake.
The experience has taught him that no cloud platform operator wants even a single outage.
“The goal is to try to get to zero,” Perry said. “The reality is, you’ll never get to zero, but, while you’re trying to get zero, the other thing, as I said, is focus on reducing the total number of customers impacted and the length of time. And I think that’s the most important thing for us to focus on.” ®