Kakao, Data Center Fire, the Data Residency Dilemma(interconnect.substack.com)
interconnect.substack.com
Kakao, Data Center Fire, the Data Residency Dilemma
https://interconnect.substack.com/p/kakao-data-center-fire-the-data-residency
7 comments
Erm, i don't know...if I were architecting the backup locations (yes, plural) for that data center in question, even if I can't reach some ideal long disatance (like the 1,100 miles mentioned), i would still opt for standing up a backup location - even if it was only a few hundred miles away...something not-so-ideal is still better than nothing. Or, maybe there were backup locations, but you know, someone failed to go through their restore/disaster recovery at other location procedures and processes? If i were the leadership, i'd be asking all sorts of questions beginning with: who cares if the backup is not 1,100 miles away, why didn't you at least have some/any back up location1?!
Microsoft Azure doesn't agree with this author. They have 3 Availability Zones in their Korea Central region.
It seems like there are a large number of commercial data centres in operation in South Korea, and not all in Seoul: https://www.datacenterdynamics.com/en/news/?tag=south-korea
Kakao's incident was caused by a single building fire. It's quite a leap to suggest it is impossible to deliver a resilient service in a country the size of South Korea.
It seems like there are a large number of commercial data centres in operation in South Korea, and not all in Seoul: https://www.datacenterdynamics.com/en/news/?tag=south-korea
Kakao's incident was caused by a single building fire. It's quite a leap to suggest it is impossible to deliver a resilient service in a country the size of South Korea.
Yeah, it's all nonsense.
Pretty much all of AWS Sydney data centers are within 100mi of each other,probably all Sydney data centers are, the entire city is in fact.
Plenty of geographical, power and environmental diversity there tho.
Given the latency differences there's a bunch of replication approaches that are practical at 5-50mi that just aren't at 1000 too
Pretty much all of AWS Sydney data centers are within 100mi of each other,probably all Sydney data centers are, the entire city is in fact.
Plenty of geographical, power and environmental diversity there tho.
Given the latency differences there's a bunch of replication approaches that are practical at 5-50mi that just aren't at 1000 too
Data center locations are chosen primarily for latency and cost (mostly power) reasons, not for disaster recovery reasons. A second datacenter near Busan would work fine.
Disaster recovery planning for "what if something takes out Seoul" obviously requires backups *outside* of the country.
Disaster recovery planning for "what if something takes out Seoul" obviously requires backups *outside* of the country.
Kinda funny the gov is calling Kakao a monopoly, yet not the chaebols?
They need some AWS/Google/Azure, screw OVH /s