Hi Han, Can you give us some more information? Where is the delay? Is it in volume building, allocation selection? Spawning? Etc. The VM instantiation process has a number of us processes that can all slow down at scale for different reasons. I doubt it is the actual spawning or allocation stage, likely a ceph/cinder building stage, especially if it’s copying data via glance to a volume. Thanks, Karl. -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://lists.openstack.org/pipermail/openstack-discuss/attachments/20230805/254e6dd8/attachment.htm>