Start a new topic

Streaming a Virtual Machine from S3

Wanted to demonstrate the streaming capabilities of LucidLink by directly streaming a VM from the cloud, to a local host. We begin by uploading a VM to our Lucid cloud instance, backended by a Amazon S3 bucket.

The VM is a 133GB file residing in a S3 Sydney (Australia) object storage bucket. My local host is a Hyper-V physical server located in Melbourne (Australia). 1st we join the cloud, we create a VM and attach the VHD located in the Lucid cloud (S3 bucket) and turn on the VM.

Voilà, VM streamed from the cloud:

As a follow-up here is what happens once the VM has been streamed already. We have cached locally the important elements of data that are required to boot the VM, therefore we don't need to download those again, we can read them locally.

So as an added benefit, you have a VM running locally which has its source from an S3 bucket, acting and behaving as if it is local.

(the reboot):


David, glad you are having fun trying to figure out ways to break LucidLink! ;)

Seriously though, this highlights a really important concept - "time to file". You said this VHD was 133GB, yet you were able to boot the machine in 13 minutes. Obviously Hyper-V did not have to download the entire 133GB in order to use it. That turns out  to be a good general rule of thumb that we are finding - the larger the file size, the better the app is written to deal with parts and chunks of it rather than requiring the entire thing to be downloaded.

Can you imagine using any other way to share a VHD across clouds the way you have done? it would require downloading the entire 133GB, which would take several hours. So the ability to go from 180 min to 13 min in your (rather extreme) test is a huge improvement.

Login or Signup to post a comment