It's an incredibly large problem. From NBO, pings to any AWS DC were about 200ms as of 2017. I think Azure and GCP were also worse at the time. They now have a

It's an incredibly large problem. From NBO, pings to any AWS DC were about 200ms as of 2017. I think Azure and GCP were also worse at the time. They now have a pop in Nairobi so if you're using accelerator it might be a lot better now, but still a huge challenge. There's also in part a trust issue. So many suppliers are unreliable, the biggest firms end up doing a ton in house.