Welcome to the Construction, Energy & Engineering SIG

LWG Webinar: Investigating Hardware Failure in Data Centres

Date: 27 March 2012 3:00 pm GMT

Description:

Equipment in data centres fail all the time and they will continue to do so. In 2008, Google fellow Jeff Dean offered an overview of Google’s data centre operations at a weekly conference for developers where he discussed failure rates within the clusters of 1,800 servers that Google uses as the building block for its infrastructure.

In each cluster’s first year, typical failures include:

• One Thousand (1,000) individual machines,
• Thousands of hard drives,
• One (1) power distribution unit, bringing down 500 to 1,000 machines for about 6 hours,
• Twenty (20) racks, each time causing 40 to 80 machines to vanish from the network, Five (5) racks will “go wonky,” with half their network packets missing in action,
• The cluster will have to be rewired once, affecting 5 percent of the machines at any given moment over a 2-day span,
• There’s about a 50 percent chance that the cluster will overheat, taking down most of the servers in less than 5 minutes and taking 1 to 2 days to recover.

Hardware and software failures in data centres cause severe interruption to businesses and their ability to function. Identifying the cause of the failure is pertinent for restoring business operations, and resolving insurance claims.

This webinar discusses the investigation techniques and methodology to determine the exact cause of hardware/software failure. The webinar is designed for loss adjusters and claims’ handlers to provide the technical background necessary to manage a forensic engineering investigation into cause. The webinar will answer questions such as:

• How does hardware fail?
• How do you investigate hardware failure?
• What was the root cause of failure?
• What factors contributed to the failure?
• Who is responsible for the failure?
• Testing Results and Determination

Register Online