As a result of a power outage, a data center of Amazon in Dublin, in the night from Sunday to Monday without electricity sat. As a result of this, many sites on Amazons Elastic Cloud service to be hosted offline.
The power outage occurred after a transformatorhuisje of the data center was hit by lightning. By the impact created an explosion, followed by fire. Although a loss of power would ordinarily need to be repacked by an emergency generator, it was found that the impact was so powerful that even the control system of the power stages was paralysed. This system ensures that the emergency generator is synchronized. Without this fasesynchronisatie can generators not be enabled automatically, so a manual action was required.
According to Amazon resulted in the disturbance Sunday night around 20 hours and the consequences were Monday morning still not entirely resolved. By the interference were both the Elastic Compute Cloud and its Relational Database Service of Amazon offline. Although the power in the datacenter within a few hours was repaired, it turned out that the back online of the various instances lasted longer than I thought. A part of the instances came again naturally online, but the remaining part had helped. Around half past seven Monday morning it was 75 percent of the instances of the so-called Availability Zone of Amazon is available again. At the time of publication were the consequences of the failure are not yet widely addressed.
Amazon has the conscious data centre in Dublin in 2008 put into use. According to the company, would this mainly European firms that no longer were appropriate in a datacenter in the US, if Amazon wanted to go. A location in Europe would lead not only to lower response times, but may also in some cases be necessary; for example, because data Europe are not allowed to leave. Microsoft has for the same reason, a data center in Dublin, which is reportedly also affected by the power outage. This would allow BPOS customers have problems. Microsoft has in Amsterdam is also a datacenter, which enables customers in the Netherlands are probably not necessarily, have encountered.
Although many companies the uptime of the cloud services from Amazon and similar providers prices, this is the second disruption at Amazon in a short time. In april this year, the provider also already been affected by a significant downtime of its cloud services. The problems with the Elastic Compute Cloud platform, Amazon would then have been caused by a faulty upgrade process, probably the result of a human error.
After the failure in april, Amazon apologized to its customers. Also promised the company that customers, who by the fault are affected get compensation; they may have ten days free of charge use of Amazons cloud services. It is, however, still the question of whether Amazon at the failure of last night’s boetekleed attracts. The company could argue that it does all it can to the uptime guarantee, but that by force majeure, weather conditions, little is to do.
Customers who are Amazon on the basis of the service level agreements (slas), want to appeal to its responsibilities, may also be a rude awakening. Amazon has as a condition in its TURN recognized that, in order to be eligible for the compensation scheme, the interference of various Availability Zones must have affected the beschikbaarbeid on a yearly basis below the 99.95 percent should have failed, which equates to over 4 hours of downtime. Although the jam is in a datacenter occurred, the facility in Dublin about three of such Availability Zones. It is not known whether all three of these areas simultaneously and long-term unavailable. Amazon recommends, moreover, to have a failover in a data center in a different geographic region to downtime as a result of a failure in a different region to limit, all of this means for many companies a significant increase of the costs.