This seems like a pretty catastrophic failure on Amazon’s part. It seems pretty incompetent actually. You should have multiple copies of your customer’s data across multiple locations.
And these multiple locations should be in different parts of the world. Not concentrated in one highly volatile region.
All this seems like common sense to me honestly. Anyone with more knowledge than me know why Amazon didn’t do this?
This seems like a pretty catastrophic failure on Amazon’s part. It seems pretty incompetent actually. You should have multiple copies of your customer’s data across multiple locations
How much of it is due to Amazon VS due to regulatory reasons (have to host data in particular region) VS customer self-serve have to choose redundancy across regions?
My quick read through the article didn’t surface an answer to this, but it is very possible I missed it.
AWS go to great lengths to explain what Az replication Vs geo replication means in clouds. "All the AZ zones get knocked out" is down to the end user to decide if they want to pay the extra for geo replication and do the paperwork etc to address that risk.
This will definitely surface which people were serious when they said "YAGNI"
And these multiple locations should be in different parts of the world. Not concentrated in one highly volatile region
AWS can't make that call unilaterally of copying people's data to other jurisdictions while also keeping to promises of data hosting and legal sovereignty. That aside in the future there'd be some horror that takes out n regions and we'd be back here saying "why oh why didn't they replicate to n+1 regions"
It's definitely a long tail event of full data loss/the kind of thing people make jokes about over paranoid resilience planning "what if someone nukes Virginia though?" but it's 99.99999% likely in the Ts & C's
This seems like a pretty catastrophic failure on Amazon’s part. It seems pretty incompetent actually. You should have multiple copies of your customer’s data across multiple locations.
And these multiple locations should be in different parts of the world. Not concentrated in one highly volatile region.
All this seems like common sense to me honestly. Anyone with more knowledge than me know why Amazon didn’t do this?
How much of it is due to Amazon VS due to regulatory reasons (have to host data in particular region) VS customer self-serve have to choose redundancy across regions?
My quick read through the article didn’t surface an answer to this, but it is very possible I missed it.
AWS go to great lengths to explain what Az replication Vs geo replication means in clouds. "All the AZ zones get knocked out" is down to the end user to decide if they want to pay the extra for geo replication and do the paperwork etc to address that risk.
This will definitely surface which people were serious when they said "YAGNI"
AWS can't make that call unilaterally of copying people's data to other jurisdictions while also keeping to promises of data hosting and legal sovereignty. That aside in the future there'd be some horror that takes out n regions and we'd be back here saying "why oh why didn't they replicate to n+1 regions"
It's definitely a long tail event of full data loss/the kind of thing people make jokes about over paranoid resilience planning "what if someone nukes Virginia though?" but it's 99.99999% likely in the Ts & C's
multi region data distribution is available for customers, but it costs more
I put this on the users
So I literally have a better backup system at home than Amazon?