No Single-Point-Of-Failure (SPOF)
Hadoop has a famous single-point-of-failure problem that is being addressed by the community. We’re taking a different approach: prepare multiple Hadoop clusters across multiple data centers.
If a Hadoop cluster dies, your job will be automatically reassigned to another live cluster. Our computing resources are always shared across our users.
Last modified: May 07 2012 04:29:02 UTC