The global BGP routing table is about to reach 768K, threatening networks with ill-prepared infrastructures. Up-to-date infrastructural hardware and timely router maintenance would safeguard systems against downtime, says Heficed, an IP address solution provider.
According to experts, next plausible internet emergency, the 768K Day, might be upon us anytime now.
There is a risk that when the global BGP routing table exceeds 768K, older routers might crash, causing a ripple effect leading to affected network’s downtime.
However, according to IP address solution providers, preventive measures or updated hardware would keep the systems unaffected.
Throughout the history of the internet, several crises were predicted to happen and cause danger to the web or computer systems.
Y2K, or the Millenium bug, was much discussed yet well prepared for, and very few system failures occurred. Other crises, like the 512K Day, caught network administrators and ISPs unprepared, causing significant troubles.
“Back in 2014, the 512K Day happened because Tier 1 ISPs, not to mention smaller providers, sleep-walked into it,“ commented Vincentas Grinius, CEO of Heficed.
“It‘s hard to imagine major ISPs repeating the same mistakes and internet being affected the way it was during the 512K, however, some serious precautions need to be taken to prevent network crash.“
The 512K Day happened in 2014 when the internet routing table for IPv4 eclipsed 512.000 BGP routes.
Most ISPs and other autonomous systems had their routers‘ memory preconfigured to store up to 512K route entries.
Once the number was exceeded, some older routers went through memory overflows and crashed.
Finally, the packet loss created by the crashes caused outages across the internet. As a result, several ISPs across the globe were affected or went down, causing them significant financial damages.
To fix the issue, network engineers reconfigured the routers to store more entries, and in many instances, the limit was upped to 768,000, hence the 768K Day.
At the time, the solution had to be applied instantly, even though it might lead to similar problems in the future.
Grinius added, “The data might differ from region to region, from one ISP to another, but we are either over the 768K limit already or will be soon enough.
“If the internet as a whole won’t be affected, ill-prepared enterprises are still at risk. Network engineers working with older routers should, therefore, re-check the BGP tables on their routers, and, if needed, optimise.
“The best solution, of course, is to update the hardware itself. The risk is real for companies operating busy networks yet not paying close attention to their infrastructural needs.
“For example, a local ISP with fairly old hardware or a non-IT company employing significant traffic should take 768K Day seriously.”
Nevertheless, per other experts, the problems might reappear anytime, even though they are predicted to be of a much smaller scale.
Others are sure that the threat of the 768K Day is being blown out of proportion. Regardless, the number of entries might already be over 768K.
“The 768K Day might happen, but I hardly imagine it being anything close to the outages of 2014.
“The providers, including Heficed, have updated their hardware to routers not possessing the risks and limitations of those that caused problems in the past.
“Even if the number of BGP routes exceeds 768 thousand by a great deal, the internet is ready,” finished Grinius.
It is yet to be seen how severe will be the problems caused by the 768K Day. In any case, taking precautions costs less than dealing with issues head-on, and most of the providers appear to have learned their lesson in 2014.