Seven-Hour Salesforce Outage Disrupts Global Services

Salesforce recently resolved a major service disruption that knocked out user access and delayed operations for roughly seven and a half hours. The outage began at 3:50 a.m. EDT on September 16, affecting multiple regions and instances, including the company's Hyperforce infrastructure. Salesforce traced the issue to an internal login service that consumed excessive server resources, which stalled requests and overloaded core system components. Although the company initially pointed to an external dependency failure on a legacy login server, it later clarified that third-party infrastructure was not at fault.

To resolve the issue, Salesforce blocked application programming interface (API) endpoints, initiated rolling restarts, and deployed regional fixes. While some automated recovery efforts succeeded, others failed, requiring manual intervention to restart instances for scheduled jobs that did not run. Most customers saw services return by 11:00 a.m. EDT, and mitigations were active across all instances by 11:39 a.m. EDT. The company officially marked the incident resolved at 2:59 p.m. EDT, promising a thorough investigation into the technical trigger.

For enterprise IT practitioners, this incident underscores that cloud migration does not eliminate architectural dependencies. Abbas Jaffery, a principal advisory director at Info-Tech Research Group, warned that such disruptions create a "temporal data problem" where delayed workflows and failed integrations cause data divergence. Jaffery advised organizations to immediately enter a reconciliation phase to verify the integrity of APIs, scheduled jobs, and downstream systems rather than assuming everything is normal once login access returns.

Security experts also noted the broader operational context of the crash, which occurred during Salesforce's annual Dreamforce conference. David Shipley, CEO of Beauceron Security, suggested the event resembled a failed software update, drawing comparisons to a December 2025 outage in China caused by Amazon's Kiro AI coding agent. Shipley also noted that recent Salesforce layoffs could have hindered the speed of the recovery process.