AKASHI Takahiro via Stratos-dev stratos-dev@op-lists.linaro.org writes:

Hi All,

# Some of you might have received a similar message, the purpose # is the same, but for wider audience.

I have been thinking of some idea of creating hypervisor-agnostic framework which will enable us to implement virtio device (backend) VMs on whatever the underlying hypervisor is. The aim is to create VMs as bare-metal applications using a small number of "common" vm services. I name this framework "virtio proxy"[1]. (You can think it as my solution proposal against the topic which Alex raised a concern on yesterday.)

Now it is the time that we have to decide whether this approach can appeal to you and meet your requirements, and so if it is worth my continuing this study in the *next development cycle* which will start in September (or October).

1. Please give me your insights/feedback against my proposal.

If we see no positive feedback or interest in the next two weeks or so, especially, from member company engineers, this study will be automatically *cancelled*.

2. If there is any interest seen, I will like to set up a dedicated

Stratos call meeting in two weeks timeframe to discuss more. What is the best time slot for you? As I live in Japan, UTC-6, 7, or 8 (or even earlier) would be the best, but I would like to hear from member company engineers in West coast of USA, the timeslot is quite flexible.

Here is my draft proposal: [1] https://docs.google.com/presentation/d/1jAOKbQpv44Rje74OI4pNKXlsCfRCQf547lHT...

I think it would be useful to discuss this at the next Stratos meeting at the beginning of September as we discuss the ways forward for the next cycle of development.

More broadly now we have a bunch of backends implemented it's time to start considering the various approaches to hypervisor agnosticism.

Mike,

Can we shift the Sept meeting to a morning slot to better align with JST?

Thanks, -Takahiro Akashi