Why Did Didi’s App Crash? A Quick Look at the Outage and Its Possible Causes

Event Recap

On November 27, users across many Chinese cities reported that the Didi Ride‑hailing app was completely unusable: location data did not display, maps failed to load, fare deductions were abnormal, and rides could not be booked. Didi’s official account apologized, citing a system fault and promising emergency fixes. By the early morning of November 28, the company announced that core ride‑hailing services had been restored, while bike‑sharing services were still being repaired and unlocked bikes were offered for free rides.

Possible Causes

Didi has not disclosed the root cause. Interviews with industry insiders suggest two main hypotheses: (1) an external attack that breached the system, and (2) inherent defects in the underlying infrastructure. The author, a former Didi employee, believes an external attack is unlikely and that a failure of foundational services is more probable. Some colleagues also linked the incident to the company’s recent “cost‑reduction and efficiency‑enhancement” initiatives.

Cost‑Reduction and Layoffs

From the end of 2021 to the end of 2022, Didi’s headcount fell from 24,396 to 20,870, a reduction of 3,526 employees (14.4%). The proportion of R&D staff, which accounts for about 40% of the workforce, dropped by roughly 1,090 people in 2022 alone. Other analysts estimate that by 2023 the R&D team may have shrunk by an additional 1,200‑1,400 members.

According to Didi’s Q3 2023 financial report, the Chinese ride‑hailing business generated a quarterly transaction volume of 72.5 billion CNY, with a daily order count of about 31.3 million. The author estimates that the outage, lasting many hours, could have caused a loss of over ten million orders and more than 400 million CNY in transaction value.

Personal View on the Outage

The author argues that beyond pure technical failure, the incident is likely tied to the company’s cost‑cutting measures, which include not only layoffs but also reductions in server expenses. Historically, Didi’s search service ran on hundreds of physical machines across dual data centers, incurring high hardware costs. Around mid‑2018, Didi began trimming physical servers and gradually migrated services to the cloud to lower costs.

During a traffic surge—such as when users repeatedly retry failed requests—the reduced‑capacity services may become overwhelmed, especially during peak hours. The author notes that while the outage was not caused by a traffic spike that broke machines, the combination of high retry traffic and insufficient capacity after scaling down could explain the failure.

Finally, the author expresses that a national‑level application like Didi should not be down for 12 hours, regardless of the underlying reasons, and wishes the Didi engineering team strength in handling this P0 incident.