Does anyone know of work being done with Firecracker for HPC? https://aws.amazon.com/blogs/aws/firecracker-lightweight-virtualization-for-serverless-computing/
Being able to spin up 1000s or instances in a few milliseconds to run a parallel computation does sound interesting. I am not that familiar with Firecracker, so I dont know how it would be useful. Also there must be some limitations, so perhaps it is not optimal for HPC - ie how much access do you get to the native networking hardware. Also I suppose that the ratio of the spinup time to the length of the complete job is important - no sense in spinning something up very fast for a job which takes hours. Any other thoughts re Frecracker? J R Hearns (currently working as an HPC roughneck)
_______________________________________________ Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
