I have been trying to get some kind of handle on the additional cost of bandwidth caused by South Africa's remoteness from the USA ie: the extra cost we would expect to pay on top of network interconnect charges once we are connected to the remote backbones. The only reference I have been able to find ( not being involved in academic institutions ) is this:
http://www.safoundation.org.za/documents/12568Telecomm_web.pdf
If one looks in Box 1 on Page 22 there are some figures mentioned. So here goes with some cigarette-box calculations ( if they still sold Texans in 30 packs, I'd have a real cigarette-box! ). Some assumptions are made.
Cost of the cables = US$650,000,000
Capacity COULD be increased :
"Given current technology, and the appropriate switching infrastructure, it is possible to increase system capacity on SAT-3/WASC to 120Gbs and on SAFE to 130Gbs."
Say another US$50,000,000 to do this, and another US$50,000,000 for maintenance, and we get:
Cost of 250 Gbs bandwidth = US$750,000,000
The cable set has a "system design life of 25 years". To be generous, let us write the cables off over 15 years.
Therefore cost of 250 Gbs per annum = US$750m / 15 = US$50,000,000
Divide by 365 to get cost per day = $136,986
Multiply by R6-90 to get SA Ront per day = R945,003
Bandwidth part:
We have 250 Gb/sec
Allow 10% for headers and stuff = 225 Gb/sec
x60, x60 = 810,000 Gb/hour
Divide by 8 to get bytes = 101,250 gigabytes/hr
Times 24 to get a day = 2,430,000 gigabytes/day
Putting it together:
2,430,000 gigabytes costs R945,003 = R0-39c/gigabyte.
This is before markup. Mark up by 100% and fudge, and you get 80c/gig on top of network charges.
I do understand that the assets ( cables ) are not being sweated and are reportedly running at 40Gbs each = 80Gbs total. Take R0-39c, divide by 80, multiply by 250, and you get R1-21c cost/gig.
Add this to what you can buy a gig of traffic for in the USA, and that is what we could be paying.
If elephants could fly, of course.
Looks so far off R75/Gb, that I must have missed something or cocked something up. Any comments?
http://www.safoundation.org.za/documents/12568Telecomm_web.pdf
If one looks in Box 1 on Page 22 there are some figures mentioned. So here goes with some cigarette-box calculations ( if they still sold Texans in 30 packs, I'd have a real cigarette-box! ). Some assumptions are made.
Cost of the cables = US$650,000,000
Capacity COULD be increased :
"Given current technology, and the appropriate switching infrastructure, it is possible to increase system capacity on SAT-3/WASC to 120Gbs and on SAFE to 130Gbs."
Say another US$50,000,000 to do this, and another US$50,000,000 for maintenance, and we get:
Cost of 250 Gbs bandwidth = US$750,000,000
The cable set has a "system design life of 25 years". To be generous, let us write the cables off over 15 years.
Therefore cost of 250 Gbs per annum = US$750m / 15 = US$50,000,000
Divide by 365 to get cost per day = $136,986
Multiply by R6-90 to get SA Ront per day = R945,003
Bandwidth part:
We have 250 Gb/sec
Allow 10% for headers and stuff = 225 Gb/sec
x60, x60 = 810,000 Gb/hour
Divide by 8 to get bytes = 101,250 gigabytes/hr
Times 24 to get a day = 2,430,000 gigabytes/day
Putting it together:
2,430,000 gigabytes costs R945,003 = R0-39c/gigabyte.
This is before markup. Mark up by 100% and fudge, and you get 80c/gig on top of network charges.
I do understand that the assets ( cables ) are not being sweated and are reportedly running at 40Gbs each = 80Gbs total. Take R0-39c, divide by 80, multiply by 250, and you get R1-21c cost/gig.
Add this to what you can buy a gig of traffic for in the USA, and that is what we could be paying.
If elephants could fly, of course.
Looks so far off R75/Gb, that I must have missed something or cocked something up. Any comments?