> On 1/19/15 12:21 PM, Nikhil Komawar wrote: > It would be great to open up a BP/spec on this. I've opened a set of BPs / specs which capture my current thoughts in the image replication space: * Glance support for multiple stores of the same type / scheme. Think cinder multi-backend support without the service per backend and scheduler. See [1] and [2]. * Glance async task to initiate image replication between 2 glance stores. See [3] and [4]. * Glance async task to initiate image location deletion on a per location basis or all locations for an image. See [5] and [6]. My thoughts here are that these BPs pave the way to a number of image based use cases related to image fault tolerance, locality and policy based hosting. Assuming we can reach an agreement on the BPs, (IMO) this is not Kilo based work given where we are in the dev schedule. Rather I suspect this work would have to happen in the L cycle... Assuming I'm in the same role I'm likely to have bandwidth to get these BPs implemented. Perhaps even a PoC in Kilo. Let's start the discussion via spec review comments and take it from there; I can attend glance meetings as needed. Thanks! [1] https://blueprints.launchpad.net/glance/+spec/multi-store [2] https://review.openstack.org/#/c/150967/ [3] https://blueprints.launchpad.net/glance/+spec/image-bits-replication [4] https://review.openstack.org/#/c/150971/ [5] https://blueprints.launchpad.net/glance/+spec/async-image-deletion [6] https://review.openstack.org/#/c/150986/