A fault in the electrical system has knocked a Level 3 data center in London offline on 10 July. The outage likely caused downtime for many companies as the facility offers space for hosting companies and provides connectivity to an Internet backbone, according to the Register.

The data center’s technicians found four "scorched" bus bars, according to an incident report obtained by the Register. Initial reports attributed the outage to an uninterruptible power supply (UPS) system fault, however, an investigation found that the fault occurred outside of the UPS.

Kevin Chapman, sales director at Piler UK Ltd. (the data center's UPS supplyer), told DatacenterDynamics FOCUS that the failure was attributable to "electrical bus-bar components external to the Piller UPS product."

The UPS units "remained stable for the duration of the fault and continued to operate as per their design, providing support to the electrical load for the duration of the autonomy level of the installed battery system," Chapman said.

The outage started around 3:30am GMT, and technicians restored power by 9am.

"A Piller service engineer provided assistance in the identification and isolation of the fault after which time supplies were restored," Chapman said.

One of the affected customers was LCHost, a web-hosting, dedicated-servers and colocation provider that uses the Level 3 data center to connect to the BT network.

“There is an outage for all ADSL customers at present due to the power failure in Level 3 Braham Street,” LCHost wrote on its Facebook page. “As a result, ADSL customers will be unable to connect at this time.”

Level 3’s Braham St. data center in London has been operational since 1999. There are five floors providing about 87,000 sq ft of technical space space total and 4MVA of site power.

In addition to BT and Level 3 itself, carriers like Colt, C&W, WorldCom, Neos, Viatel, Fibernet and 51 Degrees provide connectivity at the site.

There are four 625kVA hybrid rotary UPS modules, set up in an N+1 configuration.

In November 2011, an air-conditioning failure caused another Level 3 data center in Nanterre, France, to overheat, pushing the technicians to start shutting down servers, which brought down customer websites. The company said the issue was caused by a broken water pipe.

Note: the original version of this article has been changed to reflect the following correction. The original version said the electrical fault was in the UPS system. As we learned, that was not the case. DatacenterDynamics regrets the error.