Failure of a Cisco switch at the Newark, N.J., data center of the colocation, hosting and managed services provider Hosting.com caused intermittent network connectivity that lasted for more than 1.5 hours on Tuesday evening. The outages affected a number of businesses using services of the facility, including Amazon Web Services, Rackspace and Peer 1, according a report by Apparent Networks, a company that monitors performance of cloud computing service providers.
Apparent’s systems started detecting connectivity issues at 6:45 pm EDT. The company said the outages ended around 8:30 pm. A post on the Hosting.com Twitter feed indicated that after a dedicated switch had failed, the second failover switch had crashed as well.
After several hours of working together with Cisco staff, who suspected the problem was caused by a software bug, as the vendor's Catalyst 6509 switches crashed repeatedly, connectivity was stabilized. Hosting.com representatives could not be reached for comment in time for publication.
The provider continued investigation of the crash on Wednesday morning, attempting to replicate its circumstances in a lab, according to a post on Twitter.
The cloud services that were affected by the outage included Peer 1, Hostway, Rackspace, Verio and Amazon Web Services. According to Apparent, each of them lost connectivity for more than one hour at various locations around the US.
The information was gathered from the infrastructure of Apparent’s PathView Cloud, which tests uptime of the major public cloud providers in real time.