Power Outage Affects HostDime Data Center
Thousands of web sites hosted by Orlando-area service provider HostDime were offline this morning because of power issues in one of the company’s data centers. HostDime specializes in reseller hosting, including several hundred thousand sites hosted by its Surpass Hosting unit. Kayla Fleming of Surpass reports that the data center lost power at about 8 a.m. Eastern time due to a local utility outage.
“What happens when there is power loss is that we immediately go into UPS for a few minutes while the generator powers up,” Fleming said in her advisory to customers. “This happened normally but approximately 40% of the servers tripped during this process. Power loss and going into UPS followed by generator power has happened a few times in the past, but we have never experienced what has happened today – the power surge during this outage was extremely high.”
Fleming said that all affected servers were back online by shortly after 11 a.m. Here’s her update:
We are meeting with our electrical engineers as we speak to analyze the situation to learn exactly what happened. We still have the data center running on generator as the power in the city could still be unstable and we do not want to risk any further outages. As many of you know these kind of issues do not happen often and we are taking this incident very seriously. We apologize for the inconveniences as we know this has affected a lot of you and your clients. We will be compensating per our SLA and will ensure we find out in detail what happened, resolve it completely and put measures in place so it doesn’t repeat again.
They were up for a while, now back down. No response from HD?
Alan FiggisPosted May 27th, 2008
What hostdime don’t make clear is that their UPS system totally failed on them.
They didn’t have a redundant UPS backup arrangement to kick in so it was far from ideal.
Sites were down for a minimum 5 Hours to the full day in some cases.
There are facts still to come to light on this I think.
EdPosted May 27th, 2008
Sounds like and oversubscription issue and lack of good capacity management!
ReaderPosted May 30th, 2008
All explained here: