51 lines
2.4 KiB
Plaintext
51 lines
2.4 KiB
Plaintext
Hi all,
|
|
|
|
I nominated myself to be a candidate of Zun PTL. As the founder of this
|
|
project, it is my honor to work with all of you to build an innovative
|
|
OpenStack container service.
|
|
|
|
OpenStack provides a full-featured data center management solution which
|
|
includes multi-tenant security, networking, storage, management and monitoring,
|
|
and more. All theses services are needed regardless of whether containers,
|
|
virtual machines, or baremetal servers are being used [1]. In this context,
|
|
Zun's role is to bring prevailing container technologies to OpenStack and
|
|
enable the reuse of existing infrastructure services for containers.
|
|
Eventually, different container technologies should be easily accessible by
|
|
cloud consumers, which is a goal Zun is contributing to.
|
|
|
|
Since April 2016, in which the project was founded, the Zun team has been
|
|
working hard to achieve its mission. We managed to delivere most of the
|
|
important features includes:
|
|
* A full-featured container API.
|
|
* A docker driver that serves as reference implementation.
|
|
* Neutron integration via Kuryr-libnetwork.
|
|
* Two image drivers: Docker Registry (i.e. Docker Hub) and Glance.
|
|
* Multi-tenancy: Containers are isolated by Keystone projects.
|
|
* Horizon integration.
|
|
* OpenStack Client integration.
|
|
* Heat integration.
|
|
|
|
By looking ahead to Queens, I would suggest the Zun team to focus on the
|
|
followings:
|
|
* NFV: Containerized NFV workload is emerging and we wants to adapt this trend.
|
|
* Containers-on-VMs: Provide an option to auto-provision VMs for containers.
|
|
This is for use cases that containers need to be strongly isolated by VMs.
|
|
* Cinder integration: Leverage Cinder for providing data volume for containers.
|
|
* Alternative container runtime: Introduce a second container runtime as a
|
|
Docker alternative.
|
|
* Capsule API: Pack multiple containers into a managed unit.
|
|
|
|
Beyond Pike, I would estimate Zun to move toward the following directions:
|
|
* Kubernetes: Kubernetes is probably the most popluar containers orchestration
|
|
tool, but there are still some gaps that prevent Kubernetes to work well with
|
|
OpenStack. I think Zun might be able to help to reduce the gaps. We could
|
|
explore integration options for Kubernetes to make OpenStack more appealing
|
|
for cloud-native users.
|
|
* Placement API: Nova team is working to split its scheduler out and Zun would
|
|
like to leverage this new service if appropriate.
|
|
|
|
[1] https://www.openstack.org/assets/pdf-downloads/Containers-and-OpenStack.pdf
|
|
|
|
Best regards,
|
|
Hongbin
|