The familiar "Blue Screen of Death" (BSOD) is trending again, but this time it has directly caused a massive global impact, such as grounded flights and closed banks...
Following the outage, the change in the number of U.S. flights looked like this.
There are already countless other examples online, so I will not list them one by one.
The cause, of course, became clear very quickly: a failure during an update by CrowdStrike, which provides third-party network services for Microsoft Azure. However, in the solutions provided so far, users are required to resolve it manually.
But first, I feel that for such a massive "disaster," the problem does not lie solely with CrowdStrike; a bug in a file update is actually very hard to avoid. However, as long as Microsoft continues its current product and service strategy, the disaster of the past 24 hours will not be the last, and it certainly won't be the largest in history.
In fact, Microsoft should only take half the "blame." The other half should be borne by the IT departments of large enterprises: Why are corporate IT expenditures rising while dependency on third-party IT grows, to the point where they lose operational capacity immediately without these third-party companies?
For various reasons, public articles need to maintain a harmonious tone, but some points are already well-known. I will briefly list them:
- Windows' underlying code is poor. Even with Windows 12 approaching, they still dare not make major changes to the most fundamental architecture;
- There is no need for much discussion regarding Windows security. The practice of frequently installing a series of updates (including hardware drivers, third-party services, security patches, etc.) is itself the greatest source of security vulnerabilities and instability;
- Actually, macOS isn't much better; Apple updates also frequently reveal security flaws. However, Apple manages its products uniformly. While it uses many third-party service components, they are presented through Apple's unified gateway, unlike Windows where "anyone goes";
- Microsoft's best products in recent years were Office, and later Teams. These two products are the fundamental reason for its firm grasp of desktop market share and the massive corporate procurement of Microsoft Cloud services. As IT departments of large enterprises become more bureaucratic, preferring a "prime contractor" (like IBM?) to solve all problems, these contractors prioritize safety and source each sub-package from a single provider. Office + Teams has become the most important cornerstone of Microsoft's transition to cloud services, not Windows. This is why Wintel struggles with AI PCs, isn't it?
- Technically, Unix/Linux systems offer better stability and security. But as fewer people in large corporate IT departments truly understand Unix/Linux internals, and with a lack of strong service providers in the market, the impact of these "BSOD" issues becomes increasingly widespread;
- In short, we can never guarantee 100% technical reliability, but the rigidity of operating mechanisms amplifies any error from underlying services. Unfortunately, those large enterprises that have undergone internet and digital transformation have lost almost all self-reliance, except for paying ever-increasing annual IT bills;
- In the short term, during Microsoft's earnings call next week, this major "disaster," its subsequent handling, and even Microsoft's potential liability for compensation will be key focus areas. Affected large enterprises will also need to quickly re-examine their IT infrastructure and operational capabilities, improve emergency plans, and develop Plan Bs—though it likely won't be of much use;
- Perhaps this round of AI hype presents a massive opportunity for restructuring, as models are fundamentally changing the software service industry. However, I find it hard to believe that those large enterprises that have lost their capacity for self-reliance will have the chance to cope with the increasingly rapid impact of this "new life form."