Get started Bring yourself up to speed with our introductory content.

What is the role of hardware in network overlays and virtualization?

Jason Edelman explains the role of hardware in network virtualization and network overlays.

What is the role of hardware in network virtualization and network overlays?

First and foremost, the network hardware that makes up the physical network fabric needs to transport IP packets. This is a major point and will never change. Hardware will be needed to transport packets between systems in the data center regardless of whether network virtualization is being deployed.

Although many solutions for network overlays focus on connecting virtual switches through a tunneling protocol, such as Virtual Extensible LAN, hardware can also be used to terminate overlays. If virtual switches are used in a network overlay, integrating them with a hardware-based switch may still be required, with the hardware acting as a gateway. This will allow for seamless integration between the physical and virtual server environments, which is another important role for hardware in network virtualization. It should be kept in mind that special application-specific integrated circuits, or ASICs, are required to terminate network overlays on a hardware switch.

There have not been many details publicized about production network virtualization deployments. It is unknown how much impact network virtualization has on the capacity of physical servers (CPU, memory, and so forth) that terminate tens, hundreds or thousands of tunnels in a virtual switch, or what the impact could be on network performance. If the impact is great in these areas, it might end up that data centers require a network virtualization solution that has the capability to terminate network overlays in hardware in each top-of-rack switch vs. each virtual switch. In this case, hardware plays an even more integral role in a network virtualization solution.

Additionally, hardware will continue to play a critical role in gaining end-to-end visibility of network traffic across a data center that has deployed network virtualization. Imagine not being able to do a traceroute in your own enterprise data center to see every hop between two virtual machines. APIs and hardware that offers real-time visibility through programmatic agents will prove to be valuable in the data center. What's also important is that there be a back-end analytics engine that correlates the data being gathered and understands the behavior of the traffic and applications. This engine also can optimize the network, creating an open feedback loop.

Although hardware will continue to remain significant, customers should not forget to focus on the overall solution: For example, is there a feedback loop with bi-directional communication, and is it operated and managed easily? Are there integration and visibility between virtual and physical network environments, and so forth?

Not every customer environment will need all of the types of functions described, but these are just some, clearly not all, of the things to think about as data centers are being refreshed and redesigned to better enable the business.

This was last published in July 2013

Dig Deeper on Network virtualization technology

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.