Video: How OpenStack Swift Handles Hardware Failures
A couple of weeks ago we held a great meetup in San Francisco where John Dickinson, the OpenStack Swift PTL, took an in-depth look at how Swift handles hardware failures. I made John sit down and walk through it again – this time on video. It was a great overview examining the code paths for several request types.
- How does Swift handle a failed node so that future requests do not get routed to that node?
- How does does Swift use hand-off locations to ensure safety of data when there is a failed drive?
- How is a request routed when a request is made to a failed drive?
- How does Swift deal with bit-rot?
- How does Swift know when a hard-drive is full?
- How does Swift deal with a full drives?
This is an in-depth look into the internals of OpenStack Swift by a core member of the project. Take a look!