Google advises Cloud Engine customers to build redundancy in their cloud apps-- its European Google Compute Engine (GCE) datacentre got hit by not one, not two, but four consecutive lightning strikes, leading to a major power outage.
The incident caused permanent "persistent disk" (a type of virtual machine instance storage GCE hosts) data loss in the europe-west-b zone, located in St Ghislain, Belgium.
Google says on 13 August 5% of "standard persistent disks" in the zone started returning I/O failures and snapshot creation errors, a total reduced to "less than 0.000001% of the space of allocated persistent disks" by 17 August.
The search giant has automatic battery backup systems, but admits extended or repated battery drain has lead to older storage devices being more susceptible to power failure-- even if most persistent disk storage is running on newer hardware.
In addition Google says it will continue upgrading hardware, implementing disk resilience system and improving response procedures.
According to a 2013 ENISA report severe weather (such as storms and heavy snowfall) is the top cause of lengthy datacentre outage in Europe, with power and cable cuts affecting base stations, switches and mobile switching.