Addressing Silent Data Errors for Resilient Data Centers

Abstract

Silent data corruptions or silent data errors (SDEs) are the most insidious kinds of errors that can happen in computing systems. Essentially, there is a defect or an unpredicted problem inside the computing chip that was not found during manufacturing testing and that inconsistently changes computation results. The industry is struggling to profile, root cause and eliminate these errors.

This 1-hour webinar discusses the magnitude of this problem, the effects it has on the users’ trust of the system and the current best practices and available techniques to deal with it. The speakers highlight innovative mechanisms that will greatly assist in reducing the magnitude of SDEs and how these solutions can optimize datacenter resilience.

During the webinar, you will learn:

• What is the magnitude of silent data errors (SDEs)
• How does SDE impact end users
• What is the cost of detecting SDE using traditional methods
• Novel techniques for detecting and mitigating SDE
• Data center efficiencies improvement with better SDE containment

Moderated by Cliff Grossner, Vice President of Market Intelligence & Innovation at Open Compute Project (OCP), this webinar featured the following presenters:

• Ashish Nadkarni, Group Vice President & General Manager, IDC
• Rama Govindaraju, Director of Engineering, Google
• Walter Abramsohn, Director of Product Marketing, proteanTecs

This webinar was hosted and organized by the Open Compute Project (OCP).

‍

‍