If you miss mirantis's webcast (early morning), you can only mine your own information. The following are only personal translations and understandings, and the correctness is not guaranteed. Correct the error.
NOVA:
-
Added support for NFV. This is very important to the open NFV project, opnfv.org. NUMA seems to be a key optimization object. High-end servers, such as Xeon QPI, are all NUMA architectures. Libvirt needs to expose more interfaces to control the upper layer.
Virt driver guest vcpu topology configuration
Virt driver guest NUMA node placement & Topology
I/O (PCIe) based NUMA Scheduling
Virt driver large page allocation for guest Ram
Virt driver pinning guest vcpus to host pcpus
Pci-SR-IOV passthrough support for networking
Live upgrade. You can better support upgrade based on rolling up upgrade, especially to ensure the compatibility between the NOVA-network and the Nova-compute. For example, the entire cluster works normally when all nodes except the NOVA-network are upgraded to the new version. Zero-downtime upgrade is the final goal, and it is estimated that the implementation is not that easy. Judo adds version information for cross-module control and data interaction, laying the foundation for better upgrade in the future.
-
Scheduler. In the future, a new project called Gantt will be created independently. Judo tries its best to strip the code of the scheduler module from the rest of Nova. The advantage of independent scheduler is that the scheduling algorithm integrates the information obtained from neutorn and cinder for scheduling.
-
Docker. Annual hot projects. The support in Nova is naturally indispensable. Nova adds the pause/unpause API that supports docker instance, so that docker can mount the cinder device when the instance is started. In addition, tempest and CI seem to have a lot of work to do.
-
Ironic. Judo has finally moved the ironic driver code merge to the trunk. Ironic will be incubated as a formal project in the next release.
-
Database integration. RedHat has dug a master Michael Bayer to Improve the Performance of sqlalchemy and alembric. Ceilometer complained that it was worse than MongoDB...
Ceilometer
1. performance has improved a lot
2. Community reboot & QA
Omitted
Heat
1. rollback is supported. In the past, if the deployment failed, it could only be restored manually. Is it very earthy?
2. allows non-admin users (after admin authorization) to create resources.
Glance
First, the glance mission has changed. glance is "a service where users can upload and discover data assets
That are meant to be used with other services, like images for Nova and templates for heat ."
Artifacts is introduced to represent a common data set.
Https://etherpad.openstack.org/p/MetadataRepository-ArtifactRepositoryAPI
The store. Libraries code library is independent of the glance. Store library, with the goal of reusability and modularization.
Marconi
I changed the name zaqar. This is a message service middleware similar to AWS SNS.
Added storage driver to support redis.
Queues migration. Https://blueprints.launchpad.net/marconi/+spec/queue-migration
Keystone
LDAP integration. Keystone supports multiple backends at the same time. That is, users are authenticated by LDAP and service users are authenticated by local SQL dB.
Other interesting security projects: Barbican and Kite. Let's see for yourself. Http://redhatstackblog.redhat.com/2014/08/05/juno-updates-security/
Tripleo
Ha is supported. In addition, HA is the default action, even if there is only one node.
Some use heat templates for automated deployment.
Horizon
Sahara dashboard integration makes it easier to deploy hadoop clusters.
Javascript unbundling.
Finally, add the neutron I am concerned:
Distributed virtual router (DVR ). The virtual distributed Router separates the north-south and east-west traffic, so that the router will not become a single point and performance bottleneck.
Fully supports IPv6 and stateful and stateless IPv6 DHCP. Ra, slacc.
The HA of L3 router supports vrrp protocol through keepalived.
L2 Population & ARP responder.
There are too many things. Let's take a look. Https://wiki.openstack.org/wiki/NeutronJunoProjectPlan
Original article:
Http://drbacchus.com/whats-new-in-openstack-juno/
Http://blog.russellbryant.net/2014/07/07/juno-preview-for-openstack-compute-nova/
Http://blog.flaper87.com/post/juno-preview-glance-marconi/
Https://etherpad.openstack.org/p/MetadataRepository-ArtifactRepositoryAPI
Http://git.openstack.org/cgit/openstack/nova-specs/tree/specs/juno/approved
Http://redhatstackblog.redhat.com/2014/08/05/juno-updates-security/
Http://www.matthias-runge.de/2014/09/08/horizon-juno-cycle-features/
Https://openstack.redhat.com/Juno_previews
Http://goo.gl/jbL909
This article from the "Focus on Linux development" blog, please be sure to keep this source http://zhenhua2000.blog.51cto.com/3167594/1562412
Whats new in openstack Juno