Re:The Cuban CDN
by vidarh on Friday August 19, 2016 @08:03PM
(#12324352)
> Why though?
Cuba has 1/4 of the population of the Caribbean in one relatively small location.
Look at their network map [1] and you see they already cover major locations in South America, and has good coverage of the US, but Central America, Mexico and the Caribbean is a blan spot.
Looking at their worldwide location, my guess is that they pick their POPs largely based on whether they see a potential for cutting latency for their customers based on a mix of population size and the local transit situation.
In the US, it's relatively sparse. Probably because you don't gain much by increasing the number, as you can solve most latency issues within the US by buying more expensive transit from established backbone providers or peering directly with major ISPs at a relatively low number of peering points.
So lots of POPs == either high population or shitty latency over the regular network coupled with poor peering opportunities or other ways of cutting latency without putting actual servers in a rack. Or both.
Cuba is interesting - as someone has pointed out elsewhere in this thread, the plane traffic between the US and Cuba is quite substantial despite the travel limitations, because of the large Cuban exile community.
If Cuba liberalises network connections, chances are that despite the small current number of Cuban internet users, it would explode quickly, and that there would be a lot of US based businesses that might look to Cloudflare if Cloudflare could cut their latency.
Given the size of the Cuban population, it wouldn't take a large proportion of the population to make it the best contender for a first POP in the Caribbean.
My guess is that Cloudflare is looking to add a lot more POPs over the next few years anyway (EDIT: Supported by finding an announced from when they set up a POP in Kuala Lumpur: In September 2015 that was their 45th POP; today their map shows 83; if they add POPs at the same rate or faster until 2017, they'll be past running out of locations more suitable than a liberalised Cuba), and if you look at the empty spots on the map, and mark those areas with high population density and poor peering situation, I'd not at all be surprised if Cuba start looking quite good in the "if just they fix these couple of issues" (from the point of view of putting a POP in - I'm not trying to trivialise their problems) category.
Cuba has 1/4 of the population of the Caribbean in one relatively small location.
Look at their network map [1] and you see they already cover major locations in South America, and has good coverage of the US, but Central America, Mexico and the Caribbean is a blan spot.
Looking at their worldwide location, my guess is that they pick their POPs largely based on whether they see a potential for cutting latency for their customers based on a mix of population size and the local transit situation.
In the US, it's relatively sparse. Probably because you don't gain much by increasing the number, as you can solve most latency issues within the US by buying more expensive transit from established backbone providers or peering directly with major ISPs at a relatively low number of peering points.
So lots of POPs == either high population or shitty latency over the regular network coupled with poor peering opportunities or other ways of cutting latency without putting actual servers in a rack. Or both.
Cuba is interesting - as someone has pointed out elsewhere in this thread, the plane traffic between the US and Cuba is quite substantial despite the travel limitations, because of the large Cuban exile community.
If Cuba liberalises network connections, chances are that despite the small current number of Cuban internet users, it would explode quickly, and that there would be a lot of US based businesses that might look to Cloudflare if Cloudflare could cut their latency.
Given the size of the Cuban population, it wouldn't take a large proportion of the population to make it the best contender for a first POP in the Caribbean.
My guess is that Cloudflare is looking to add a lot more POPs over the next few years anyway (EDIT: Supported by finding an announced from when they set up a POP in Kuala Lumpur: In September 2015 that was their 45th POP; today their map shows 83; if they add POPs at the same rate or faster until 2017, they'll be past running out of locations more suitable than a liberalised Cuba), and if you look at the empty spots on the map, and mark those areas with high population density and poor peering situation, I'd not at all be surprised if Cuba start looking quite good in the "if just they fix these couple of issues" (from the point of view of putting a POP in - I'm not trying to trivialise their problems) category.
[1] https://www.cloudflare.com/network-map/