Martello BLOG

An Internet Outage is Closer Than You Think, How “768K Day” Will Affect Your Network

As the global routing table increases, routers are using more memory to hold these tables. Naturally, memory on network hardware has its limits and when we reach that threshold what can we expect? An event that potentially could cause significant disruptions across the Internet, the so-called “768k Day” is closer than you think. This day is anticipated to be coming this month when the global Border Gateway Protocol (BGP) routing table is expected to exceed 768,000 entries. BGP is a routing protocol used to transfer data and information between host gateways, the Internet or autonomous systems. Without it, the Internet wouldn’t work.

So why is 768K day a big deal?

Let’s rewind 5 years. In 2014, there were concerns over what has come to be known as “512k Day”. This name was given to the moment when the Internet routing table was going to exceed 512,000 BGP routes.  Many ISPs and other organizations had provisioned the size of the memory of their routers for a limit of 512K route entries. Once the route entries exceed that maximum capacity some older routers suffered memory overflows which caused their CPUs to crash. These crashes caused significant packet loss and traffic outages across the Internet, including some large provider networks. Engineers and network administrators scrambled to apply emergency firmware patches and set a new upper limit of 768k entries. Since memory is expensive, many vendors saw 768,000 as a decent trade-off for memory utilization.

512k Day was an Internet milestone much like the y2k crisis at the end of the 90s, and a wake-up call for a lot of ISP and Internet organizations. This looming issue wasn’t a secret, yet many providers were caught off-guard with outdated network equipment that had a ripple effect into Internet outages.

Fast forward 5 years later, and the upcoming  768k Day is becoming an echo of 512k Day, but with a higher threshold. This time around, things will be slightly different though. Most large providers who felt the impact of 512k Day have learned their lesson and have likely prepared and maintained their infrastructures to accommodate the next level of usage. As a result, we hope not to see the same level of outages as we did in 2014.

A lot of smaller ISPs, data centers and other providers may not be as well prepared, however. When you look at Internet paths, a good amount of service traffic goes through these “soft spots” of Internet infrastructure. These areas of the Internet where maintenance on legacy routers can be easily missed, meaning we may still experience issues or outages on 768k Day.

While BGP is the core to the functioning of the Internet, there are more flexible solutions to provide business continuity for enterprises and end users.  Link management is a technique which has been in use for a number of years, but cost and complexity have hindered many organizations’ efforts in taking advantage of this increase in productivity. BGP programming is the traditional method of achieving the result, but it is a costly approach. One aspect of BGP that users must keep in mind is that it is not designed to do link balancing, but to perform internet connectivity.

ISPs now offer managed BGP services but such services have a high price tag attached to them and can create other types of issues, related mainly to responsibility for problem resolution. BGP also has limited link management abilities forcing a specific link per traffic type and does not have the ability to dynamically redirect traffic.

Want to learn more about link balancing? Download our eBook!

So it should go without saying that if you make sure your network is up to date before 768k Day arrives then you have nothing to worry about!