'768K Day' Is As Overhyped As Y2K, ISP Says

Some experts have warned that outdated networking equipment could cause significant connectivity headaches for consumers and companies alike as part of what many are calling “768k day,” expected to occur sometime next month.

But internet service providers and network engineers contacted by Motherboard say they have a good handle on the problem, and the threat is largely being overhyped. The worry is that we’re about to see a repeat of a major outage that crippled parts of the internet back in August of 2014. Dubbed “512k day," the outage resulted in widespread connectivity and packet loss issues across hundreds of global ISPs, resulting in millions of dollars in lost productivity from industries and consumers being kicked offline. 512k day was caused by routers running out of memory for storing the global BGP routing table, a file containing the addresses of all devices and internet-connected networks worldwide. In 2014, most network hardware was capable of storing 512,000 internet routes in memory. When Verizon added 15,000 new routes on August 12, 2014, it inadvertently triggered outages that reverberated across the internet.

Many of these routers were updated with newer firmware that bumped the size of the memory allocated to handle the global BGP routing table to 768,000, or 768k. Now, a recent report by ZDNet noted, firms that track the size of the global BGP routing table say the table is pushing toward that 768k limit. A Twitter bot dubbed BGP4-Table has been tracking the internet’s approach to this limit, and indicates it should be bypassed sometime in the next month.

You’d think that this would be concerning for internet companies, but telecom executives and engineers consulted by Motherboard weren’t particularly worried about the threat. “I think it will resemble the fallout from Y2K,” Dan Jasper, CEO of independent California ISP Sonic told Motherboard in an email.

Jasper noted that updating older routers doesn’t pose much of a challenge to savvy telecom engineers, stating that “table size management and prefix filtering is pretty simple stuff.” Sonic CTO Nathan Patrick offered a little more insight to Motherboard, reiterating that the problem isn’t so much a bug as a simple physical limitation of older routers. He also was quick to note that the problem only impacts routers at the core of the internet, not residential gear. “While 768k was a relatively common value for IPv4 table size from many vendors, the equipment that had that limitation is quite old, and mostly retired from full table service at major providers,” Patrick said. “So, while a small ISP or enterprise somewhere may see an impact, I'd peg large-scale doom as very unlikely.” Verizon—which has come under fire for lagging on network upgrades in states like Pennsylvania, New Jersey, and New York—did not respond to Motherboard’s request for comment. Hardware manufacturer Cisco told Motherboard it was aware of the looming threat and noted that more modern network routers support millions of routes. The company directed Motherboard to a recently published guide for network engineers aiming to avoid disruption on 768k day. “In short, 'the sky is not falling,'” Cisco said in the guide.

Get six of our favorite Motherboard stories every day by signing up for our newsletter.