Web users may experience local connectivity problems following the passing of a recent milestone for the internet’s infrastructure. As of 2013, Internet Protocol version 4 (IPv4) was the most popular means of directing traffic across the web. IPv4 distributes IP addresses to routers, commonly referred to as “nodes,” that exist on an IP network. Nodes provide internet access to web-enabled endpoints on an IP network like mobile phones, computers, and certain gaming devices.
Currently, IPv4 is able to provide about 4.3 billion unique IP addresses to nodes globally, but as adoption rates of internet-enabled devices continue to grow exponentially, the global routing table increases in volume and IPv4 is running out of addresses that it can allocate. The Cisco 7600/6500 and RSP720/Sup720 series are some of the most popular routers currently being used by ISPs, and by default they are built to handle just over 512,000 BGP routing table entries.
The problem? The global routing table has just grown past the 512k mark by about 15,000, causing widespread network instability. Verizon is primarily responsible for the creation of the new entries that pushed the net over the edge, but in fact, the problem is not as serious as it sounds. IT analysts have been anticipating “512k Day” since the mid-1980s, and Cisco addressed its inevitability mid-May of this year.
eBay, Comcast, and Time Warner are thought to have felt the disturbance in the internet caused 512k after each provider saw hiccups in service this week. Liquid Web, an internet service provider that maintains direct connectivity for Comcast, Verizon, and Level(3), publicly attributed an outage to 512k Day on Tuesday.
We are in communication with providers to attain a timeframe for these widespread outages they're experiencing but have not gotten one…
— Liquid Web (@LiquidWeb) August 12, 2014
Most end-users have yet to experience the effects of routers crossing the 512k limit, but as more addresses are added to the global routing table, more routers could be hit. A fix, according to Cisco, is possible via a reconfiguration of an affected router’s memory allocation.