As part of the VMware NFV team, I meet service provider customers on a daily basis. I often get the question about choice of hypervisor. When I ask why the focus on the hypervisor, customers often tell me it’s because they believe the choice of hypervisor determines the potential for openness, flexibility and choice. This is not the case. Here’s why:
- You don´t deploy a hypervisor, you deploy a service
- To deploy a service, you don´t buy a hypervisor, you buy an NFV platform
- To buy an NFV platform, you don´t choose a hypervisor, you choose a platform partner
- Every platform partner has their own solution, vendor specific configurations of OpenStack, KVM, vSphere etc.
But here’s the key: platforms don’t interoperate at the hypervisor, they interoperate at the VIM (virtual infrastructure manager). And to support a successful NFV deployment your platform needs to be able to interoperate across more than one VIM
OpenStack will, in time, support cloud-ready VNFs, but many VNFs are not yet cloud ready. That means many early VNFs don’t understand that they have a flexible cloud infrastructure beneath them. To support these non cloud ready VNFs the platform must emulate dedicated hardware, not a cloud platform. That means another API (application program interface). Another VIM.
Next on the horizon are containers that promise Hyper-scale, and another VIM implementation.
So for an NFV platform to enable success for the operator, it must support multiple VIMs today, and be backed by a strategy to cater for any future evolution at the VIM layer. Regardless of which hypervisor is deployed.
That’s why VMware has developed interoperability with more VIM´s than any other vendor in the industry, and continues to invest in that strategy.
To learn more about VMware´s NFV offering please visit us here.
Henrik
