TheJournal.ie uses cookies. By continuing to browse this site you are agreeing to our use of cookies. Click here to find out more »
Dublin: 5 °C Friday 24 May, 2013

ESB denies that lightning caused last weekend’s Amazon power cut

The ESB has said that it was a transformer failure that caused the outage on Sunday. Amazon’s backup generators failed after the power cut.

Image: AP Photo/Paul Sakuma

THE ESB HAS said that a transformer failure, and not a lightning strike caused a power cut which knocked out supply to 100 customers in CityWest, including one of Amazon’s data centres on Sunday.

Silicon Republic reports that the ESB has ruled out lightning as the cause, even though Amazon had said that lightning had caused an explosion at the CityWest substation.

ESB staff had initally thought that lightning was the reason for the outage, but the company has now said that a transformer failed.

Data centres belonging to Amazon were among those affected. Silicon Republic reports that some Amazon customers were disrupted for up to 48 hours.

It’s also reported that Amazon placed a connection between the ‘explosion’ and the failure of its own backup generators.

Centres owned by BT were also hit, though in those cases back-up systems kicked in immediately and ensured no impact for customers.

Read more from Gordon Smith’s report on Silicon Republic>

Read next:

Comments (7 Comments)

  • Seems to me that Amazon has no one to blame but itself if its data center lost power…. They should have had ample UPS to bridge the utility power loss and the time their N+1 generators kick in. Also, they should have diesel tanks onsite to store at least 72hours worth to run the center on full load, and a local supplier on contract to re-fill the tanks after 24 hours. If Amazon do not have any these redundancy measures in place, I would be very surprised… If they do and they still had power loss that affected their customers services, then they need to look internally for where to point the finger of blame!

    Reply
    • Not to mention their own software hampering restoration of the downed instances and a complete lack of help from their support in trying to restore things. This has been a joke from the beginning.

      I also cannot understand why they only had power from one sub station supplying the DC. Any data centre I’ve been involved with has had power coming from at least 2 sub stations to avoid these issues.

      Reply
  • People who think that Ireland could make some sorely needed money by being a reliable cloud computing bridgehead for Europe.

    Reply
    • Loss of power is a serious issue. Hundreds of thousands of children on the brink of death by starvation with no back up systems. This is going to have a detrimental effect on their profitability.

      Reply
  • Maybe there was a lack of surge protection? 48 hours is a long time for a data center to be down. There must have been equipment damage. All data centers will have bought at least two independant feeds from the esb as well, were the esb supplying both from the same substation?

    It is very embarrassing for this to happen. Ireland has been trying to market itself as an ideal data center location because of the lack of extreme weather conditions, distance from earthquake zones and political stability. Top class electrical infrastructure is part of that marketing speil.

    We do not want to lose the investment and employment brought by these corporates.

    Reply
  • who gives a crap really ?!?!

    Reply
    • I give a crap. Our company had a good chunk of our infrastructure down
      for 2 days because of this. Im an Irish system administrator working in Spain and this was embarrassing to say the least. We chose Amazon for its reliability and really cutting edge cloud services. The fact that its in Ireland was a bonus and made me proud that one of the best cloud facilities in Europe, if not the world, is in my home country. But we’re going to think hard about continuing with Amazon as I’m sure many others will. It’s a pity but when your company’s existence depends on servers staying on, and even minutes of downtime cost thousands in revenue and incalculable loss of reputation, you’ve got to go where you’ll be sure this kind of thing doesn’t happen.

      Reply

Add New Comment