Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

That's not the bottleneck. The bottleneck is the IP interconnect capacity between edge network and its peers. If you only have 4x100 Gbit/sec IP sink to Level3 in SEA and your other sink is 100G to NTT, no amount of traffic engineering in your edge network is going to give you more than 500 Gbit/sec of exit and that exit will only be available if NTT has entire 100Gbit/sec that is not congested and Level3 has 400Gbit/sec not congested.

In eyeball networks the bottleneck is not upload. It is traffic being brought into the edge network from upstreams (netflix/photos/etc)



This is a accurate explanation. Eyeballs are inbound heavy and have plenty of outbound capacity (it's also why they don't care about outbound ddos so much...). However it's important to remember that they can easily congest their internal network infra on the inbound.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: