This post is Sponsored by – My Linked Profile
Martin started out pretty well, talking about the value proposition of NSX. I asked the question earlier how does NSX bring value to the enterprise. The simple value proposition is what gets a solution in the door. For example, the server virtualization value proposition is simple. You can save money if you consolidate your Windows & Linux workloads to fewer servers. Once server virtualization is in the door more of the operational advantages and efficiencies appear. An example is the ability to spin up a server within seconds vs. hours or days.
So, what’s the value proposition for NSX? Martin, first level set on what we are talking about when we talk NSX. NSX is basically a network hypervisor similar to how ESXi is a server hypervisor. So, NSX would sit on top of the physical network, which would give you the same capability to provision extremely complex virtual networks from Layer 2 to 7. Network virtualization allows for more powerful automation. VMworld’s day 2 keynote touched on the power of automation.
The network industry has focused on protocols for a long time because of the need interoperate hardware devices between vendors. The industry needs to move past the low level how do you implement to a services model. The method that he suggests to get there is through abstraction. Abstraction allows us to begin to think past the bits and think about how the services are provided and managed.
Two areas that he called out for high level of integration is top of rack switches. This area allows for high port density implementations to integrate with the virtual network. The Second are is the higher level layers such as load balancing. F5 for example has announced a load balancing solution that support VXLAN for load balancing hosts connected to a physical port with a virtual host connected to the virtual network.
“Fast is the new better, Fast is the new cheaper, Faster is the new faster!” – Chris Launey Director, Cloud Hosting The Walt Disney Company
The business case is all about eliminating the bottlenecks in your operations. The ability to take a similar operations model to Server Virtualization and apply it to Networks.
Martin’s argument is that all the added functionality is already happening at the application level. If it’s happening at the application level now then why continue to add functionality at the hardware layer? Virtual Server AccessPorts having been growing exponentially. This is not specific to virtualization. Simply put more workloads are running in virtual environments so the network is running within this environment.
A sample use case he gave was the physical limitations that non-virtualized network place on an environment. By virtualizing the network you can take something that was bound by physical location restrictions and place in physical somewhere else without interruption of service. Think of it as P2V or even vMotion for hosts connected to the physical network.
A common theme that he gets is that debugging the network becomes difficult. VMware addressed this in an earlier post.
This session was just as good as I expected. I expect to have some more technical (any one say recursive or nested networks?) and non-technical posts spawned from this session as I watch it again and again. Great material and thanks to Martin.