Virtual private server host Crucial Paradigm has blamed "human error" for a six-and-a-half hour outage that hit 1400 customers yesterday.
Managing director Aaron Weller said approximately 500 servers — about a quarter of its server farm — were impacted.
Apart from customers' virtual machines, Crucial's own DNS servers, and shared and reseller services, also took a hit.
The outage started roughly at 3pm AEDT yesterday and was fixed at approximately 9.30pm.
Weller apologised for the outage in a statement to iTnews, noting it was the first time its environment had been adversely impacted "to such a degree".
"This was not a technology failure but the result of human error and we are taking urgent steps this morning to ensure this kind of problem never happens again," he said, without going into specifics.
The company said on its forums that the initial fault related "to the network that provides our SAN network connectivity".
It further posted to Facebook: "The issue relates to a network configuration fault which brought down our redundant links between the pools and the SAN."
A full post-incident report is expected to be made available to customers by close-of-business Friday.
Weller said that "occasional issues are inevitable" in hosted environments, and that customers should prepare "policies and contingencies" to shield their business or customers from the flow-on impacts.
The comments came after several impacted customers criticised redundancy and restoration arrangements.
"2 hours downtime is not good. No redundancy! This is bad and unprofessional. This is the second time in 2 weeks - and last night around 2am," one posted to Facebook.