Decentralization is something that shouldn't be forgotten.
To make the platform resilient, all block producers can't use the same service provider. In an ideal world, all top 20 should use completely different systems.
Decentralization is something that shouldn't be forgotten.
To make the platform resilient, all block producers can't use the same service provider. In an ideal world, all top 20 should use completely different systems.
and of course geographically dispersed as well, especially for a blockchain!
and as STEEMIT also experienced a few months back right after switching to AMAZON's "premium" cloud earlier this year, even the "creme de la creme" of cloud services can get "dinged" as well.
From @mrosenquist's post earlier this year:
Link: The Real Lesson from the AWS Outage
And more on the failure itself:
Link: Amazon’s massive AWS outage was caused by human error... One incorrect command and the whole internet suffers.
While massive decentralization helps, it doesn't solve everything, and also brings on its own set of potential issues, which @Dan acknowledges up-front with EOS and plans to address by means of various clever and ingenious design features. That's why 99.99999% uptime and SLAs will always be an illusion. Murphy's Law works hard to make sure of it!
Link: EOS.IO Technical White Paper