The Multi-hypervisor OpenStack Cloud: Are We There Yet?
One of the visions for OpenStack is for it to be the unifying common fabric for all cloud and infrastructure components in the enterprise. Literally, this would mean that all hypervisors, storage systems, and networking services from different vendors could be offered to users and orchestrated under a single umbrella of cloud APIs.
In Paris, we gave a talk on “OpenStack and vSphere/vCenter”
and used OpenStack+VMware as an example of this vision being executed. Back then, it was clear that the main problem was networking—the options for full-featured and reasonably performant programmable networking with multiple hypervisors were limited.
In this presentation, we’ll take another stab at this problem and see what has changed in the last six months. In particular, we will discuss:
- Use cases for multi-hypervisor (aka heterogeneous) OpenStack clouds
- What are the benefits of KVM and ESXi co-existing in the same cloud?
- What can we enable with OracleVM and containers (LXC, Docker)?
- Integration points and high-level architecture for heterogeneous clouds
- Solution architectures for networking, including:
- VMWare NSX J
- uniper Contrail 2.0
- Cisco 1000v
- Demo and Q&A - are we there yet :) ?
By attending this presentation, you’ll gain a solid understanding of today’s options for using multiple hypervisors in the same OpenStack cloud, and why you may (or may not) need it in your environment.