Russia VPN caused outage

On April 3, shoppers and commuters across Russia found card terminals, mobile banking apps and ATMs failing for roughly an hour. (themoscowtimes.com) The disruption hit Sberbank, VTB and other large lenders and briefly forced the Moscow metro to open its turnstiles without contactless payment. (usnews.com) Pavel Durov, the founder of Telegram, traced the failure to state efforts to block VPNs. (bloomberg.com) He said the government’s blocking attempts “triggered a massive banking failure” and noted tens of millions of Russians use Telegram through VPNs. (bloomberg.com) Independent reporting and industry sources offered a concrete technical mechanism for how an anti-VPN operation could knock over payments. (dw.com) Russian censoring equipment sits inside operators’ networks and inspects traffic at scale using deep packet inspection (DPI). (dw.com) When the state ramps up filtering rules — to block VPN protocols, spoofed TLS handshakes, or hundreds of IP ranges — those DPI boxes must apply many more checks to every packet. (dw.com) Sources told reporters the filters can “crash under load,” dropping or delaying traffic at a chokepoint that sits upstream of banks’ mobile apps and payment gateways. (dw.com) That failure mode is simple to picture. A bank app opens a TLS session to its backend. Packets leave a phone, traverse a mobile operator, and hit carrier equipment where DPI rules run. If those devices slow or silently drop packets while they slog through rule-matching, the app’s TCP connections stall, API calls time out, and a payment instruction never reaches the acquiring network. (themoscowtimes.com) The incident shows a common blind spot in resilience planning. Firms harden application logic, replicate databases, push critical paths into FPGAs, or use kernel-bypass to shave microseconds. Those efforts protect against software bugs and server failures. They do not protect against a failure that occurs on someone else’s network device an ISP inserted between clients and your APIs. (themoscowtimes.com) Russian regulators moved quickly to control the narrative. Roskomnadzor ordered outlets to remove reports linking the outage to internet blocks. (uawire.org) Banks, meanwhile, reported that services were restored after an interruption. (akm.ru) For architects of latency-sensitive systems, the takeaway is concrete. Design assumptions must include upstream policy and filtering as failure modes. Test multi-homing not just for latency but for policy diversity. Validate that critical flows survive transparent middleboxes, packet reordering and aggressive DPI rules. Plan for out-of-band settlement or manual fallbacks the way merchants in Moscow had to accept cash. (themoscowtimes.com) The day Russian card rails hiccupped, commuters entered the metro and a zoo asked for cash while digital defenses and political controls collided inside carrier boxes. (usnews.com) Sberbank later confirmed access to services had been restored. (akm.ru)