Because I am sensitive to the fact that I am a recovering academic I am hesitant to get drawn into any “How many angels can dance on the head of a needle”-type discussions. Recently, however, I have found myself being drawn into discussions with people who associate NFV (Network Functions Virtualization) strictly with service providers. In most cases I ask these folks: If you don’t think that NFV is applicable to enterprises, what about NSV (Network Services Virtualization)?
The linkage between NFV and service providers
Just over two years ago an Industry Specifications Group (ISG) for Network Functions Virtualization (NFV ISG) was formed under the auspices of the European Telecommunications Standards Institute (ETSI NFV ISG). While the membership has evolved significantly, the initial members of the ETSI NFV ISG were all service providers such as AT&T, Deutsche Telekom and NTT.
Table 1 contains examples of functions that the ETSI NFV ISG believes can be virtualized:
Given the leadership role that ETSI is playing and given the interest that they have in virtualizing functionality such as broadband network gateways and radio network controllers, it is easy to see why some people associate NFV strictly with service providers.
The Open Networking User Group (ONUG)
ONUG was founded in 2012 and unlike ETSI its members are primarily enterprise companies such as Fidelity Investments, Citigroup and FedEx. In a white paper entitled Open Networking Challenges and Opportunities, the group discussed the cost and complexity of managing a large number of Layer 4 – 7 network appliances from different vendors with different management tools. The appliances they mentioned were:
- Server load balancers and application delivery controllers
- WAN optimization
- SSL/IPSec VPNs and Intrusion Detection and Prevention Systems
In that white paper ONUG coined the phrase Network Services Virtualization (NSV) and stated that the NSV use case “Seeks to leverage the flexibility and low costs of commodity servers to establish a scale out pooling of virtual and physical appliances, which can be put to use servicing applications.” ONUG went on to say “As each Layer 4 – 7 function is virtualized in software, it provides the following benefits:
- Lower CAPEX costs (approximately 30 percent less)
- Rapid service provisioning, ability to deploy Layer 4-7 services that follow specific virtualized sets of applications
- Reduced risk through service distribution
- Eased management and reduced operational costs; ability to be centrally managed by generalized IT operational teams
- Consistent policies across different Layer 4-7 services and across data center, campus, and WAN networks
- Programmatic control and ability to offer network functions as a service to developers
Other potential benefits of NSV include the ability for IT business leaders to deliver on-demand or self-service IT delivery to business unit managers.
Are there differences between what ETSI is trying to accomplish with NFV and what ONUG is trying to accomplish with NSV? Yes, there definitely are. Service providers hope to virtualize some functionality that few if any enterprise organizations implement and their need for scale far surpasses what is needed by the vast majority of enterprise organizations. However, if you change at most a few words in how ONUG describes the NSV use case, it sounds exactly like what ETSI and others are trying to achieve with NFV. In addition, if you look at the list of appliances mentioned in the ONUG paper, they are all contained in the table above.
The bottom line is that conceptually NFV and NSV have far more points of commonality than differences. I will leave the argument of whether or not what enterprises are attempting to do to establish a scale out pooling of virtual and physical appliances is indeed NFV or is more appropriately referred to as NSV to people who have way too much time on their hands.