3 ms·
it is also a lot more expensive to achieve true 1:1 redundancy, for a mid sized ISP. let's say you're an ISP in monrovia, liberia. you can buy for N dollars pe
by walrus01 9y ago
it is also a lot more expensive to achieve true 1:1 redundancy, for a mid sized ISP.
let's say you're an ISP in monrovia, liberia. you can buy for N dollars per month a non redundant, single 10 Gbps transport circuit via the ACE cable that comes out the other end at some carrier POP in france, non diverse.
now you want to achieve some sort of diversity in case your cable route is cut, as the present cable has been.
you need to buy another separate 10Gbps circuit from monrovia to accra, ghana, going the other direction, establish a node in accra, and then buy another (physically diverse, in a different submarine cable) 10 Gbps circuit from accra to western europe.
or buy a circuit from monrovia to cape town and then something that is guaranteed to be physically diverse from cape town onwards to the internet. either by getting a transport circuit from cape town to western europe/france/UK, or buying IP transit from .ZA ISPs, or some combination of the two.
- leoc 9y agoOnly-slightly-relevant but fun fact: back around the late ‘90s the UK’s university network JANET lost its Internet connection to the US through a combination of a cable outage and some bad preexisting dealmaking and BGP config. In the end of course it got fixed and JANET got firmly reorganized, but for several days or weeks you really could reach most or all European servers from UK universities but not US (North American?) ones. My details are fuzzy because it’s not easy to find any information about the episode anymore, but it really did happen...