Hacker News Reader

Comment by ai_ja_nai | Hacker News Reader

So Google has no redundancy at datacenter level?

Not for block storage (aka persistent disks): "we would like to take this opportunity to highlight an important reminder for our customers: GCE instances and Persistent Disks within a zone exist in a single Google datacenter and are therefore unavoidably vulnerable to datacenter-scale disasters. Customers who need maximum availability should be prepared to switch their operations to another GCE zone. For maximum durability we recommend GCE snapshots and Google Cloud Storage as resilient, geographically replicated repositories for your data."

Achieving RPO=0 generally requires synchronous replication to a different datacenter which adds significant latency.

Amazon builds datacenter redundancy in the same geographic locale. You can then setup synchronous replication between datacenters without atrocious latency, as all the disks are fairly close by, albeit powered [in an emergency] by independent generators.

OTOH, Google does datacenter redundancy across different locales, which make synchronous replication perform much worse, like you noted.

GCE appears to have a similar region/zone structure as Amazon, but neither provider replicates block storage across zones.

I stand corrected. Looking at https://cloud.google.com/compute/docs/zones?hl=en, it seems that GCE has multiple zones in the same locale (region). I might have confused with Google's internal setup, which during my tenure there, was heavy on using replication across locales, up to systems like Megastore and the 200ms synchronously replicated commit.

Reply on news.ycombinator.com