Snowflake users worldwide experienced serious problems this week after a recent update introduced errors in the data warehouse platform. The failed software change temporarily disabled important functions, preventing customers from running queries, loading data, and causing error messages.
The cause was a database schema change that proved to be backward incompatible. This created version conflicts between different parts of the Snowflake platform, leading to failed or severely delayed operations. The first problems were reported around 02:55 UTC and lasted more than 13 hours in some regions.
According to The Register, the outage had a global impact and affected ten of the twenty-three Snowflake regions. In the United States, this included Azure in Virginia and AWS in Oregon, while European and Asian regions, including Ireland, London, Zurich, Sweden, Mumbai, and Singapore, also experienced problems.
Recovery and criticism
The Register reports that Snowflake indicated it had identified the cause about an hour and a half after the first reports. In some regions, systems became available again around 05:00 UTC, while recovery took longer elsewhere and service was not immediately stable. Users complained that rolling back the update took an unnecessarily long time.
The incident follows shortly after an earlier disruption in early December, when Snowflake also experienced performance issues in one of its AWS regions. According to The Register, the outage is part of a broader series of recent problems affecting large data platforms, including competitor Databricks.
Snowflake has promised to publish a full root cause analysis in the near future. The company wants to provide more clarity about how the error occurred and what measures are being taken to prevent it from happening again.
While Snowflake typically publishes a preliminary cause and a subsequent root cause analysis during incidents, this is not the case for all competitors. According to The Register, Databricks does not typically disclose the technical causes of outages. These differences in transparency play a role in how customers evaluate cloud providers during large-scale incidents.