Last Thursday, I published a blog entitled “The 7 Myths of SDN”. Yesterday at Juniper’s Partner Conference in Las Vegas, I dispelled those myths at the beginning of a presentation called “Decoding SDN.” For those of you who haven’t had a chance to see the presentation, I thought it would be helpful to quickly outline those myths here.
Myth 1: It’s only about data center networking
Truth: SDN applies to all networking and network services -- from the enterprise data center & campus to service provider networks.
As organizations look to implement private cloud data centers, the static nature of networking presents acute challenges. Because of this pain, there is a lot of focus on using SDN to enable networks to dynamically respond to cloud orchestration systems.
But SDN will apply much more broadly than that. SDN makes sense in a large campus environment. It also will help service providers as they evolve their networks to create more capability for their customers. This is particularly true for edge networks – which is the place that service providers deliver value to customers, thus generating new revenue for the service provider.
Myth 2: It’s only about reducing CAPEX
Truth: OPEX reduction is more significant.
This is simple math. Most organizations spend a lot more money on OPEX, particularly people costs, than they do on new equipment. SDN will enable the automation of processes that are manual today, thus providing significant OPEX savings.
SDN will reduce capital costs as well but the bulk of the savings will come from automating process and thus reducing overall labor costs.
Myth 3: It’s only about software
Truth: SDN will fuel hardware innovation.
As more people do more things online, the demand for network capacity continues to grow at an astonishing pace. The complexity and manual configuration of today’s networking software puts limits on how fast network capacity can grow. SDN will help networks break free from the limits of today’s software.
You can think of network hardware as a software accelerator. When you offload something into networking hardware, you can make it run an order of magnitude faster, or more. As SDN frees networks from the limits of today’s software, the demand for faster and more capable hardware will increase. This will provide an opportunity for the entire networking industry to further innovate in hardware.
Myth 4: It’s only about centralization
Truth: Considerable intelligence stays decentralized.
Centralization is a powerful concept and is one of the key principles of SDN. Centralizing functions simplifies design and lowers operating costs. However, networks are by nature decentralized. Remember, networking is all about moving data from one place to another – which is inherently distributed.
Even as some functions become centralized with SDN, it still makes sense to keep considerable intelligence decentralized. That improves performance and makes the network more resilient to failures.
The confusion about centralization with SDN reminds me of the discussions around 2000 about the Network Computer. The idea of the Network Computer was that all of the applications moved to the center and end-user devices would be just dumb HTML terminals. It never happened. Instead, we have a whole new set of smartphones and tablets. These devices rely heavily on central cloud processing – but they are also quite capable; they run applications and they do the work of providing the graphic user interface to present the data.
The same thing will happen to network devices with SDN. Centralized functions will play a much more important role but the decentralized devices will remain intelligent and continue to provide lots of value.
Myth 5: It’s only about OpenFlow
Truth: OpenFlow is just a protocol and probably not the most important one for SDN.
As is often the case with new trends, people have gotten confused in some ways. One of these confusions is that SDN = OpenFlow. SDN is a very broad and highly impactful way of building networks; OpenFlow is just a protocol – one of hundreds that are supported across networking systems. OpenFlow will play a role in SDN, but it is not nearly as significant as some think it is.
At Juniper, we think the most important concept in SDN is “SDN Service Chaining.” This provides a way to pull services out of network and security devices and chain them together in software. SDN Service Chaining is an important architectural concept that can be broadly applied to the challenges of today’s networks. OpenFlow can’t be used to build SDN Service chains, thus my view that OpenFlow is probably not the most important SDN protocol. We will see new protocols (plus probably some extensions to existing protocols) emerge to support SDN Service Chaining.
Juniper will support OpenFlow across our switches and routers in 2013. OpenFlow is useful for some applications, but it is not a panacea and it certainly is not the same thing as SDN.
Myth 6: It’s going to happen immediately
Truth: It will happen step-by-step.
I actually think it will have a more significant, long-term impact on networks than most people realize. SDN is a major shift in the networking industry – from delivering all of the value in hardware systems to delivering much of that value in software. This means packaging, licensing, and pricing that software separate from hardware systems.
But these things take time. Customers have operating networks today; you can’t snap your fingers and magically transition to SDN. You need to introduce SDN capabilities into existing networks in logical steps, typically deployed one at a time.
Juniper has defined a four-step roadmap for customers to deploy SDN within their networks; you’ll find more about those steps in a recent blog.
Myth 7: It’s going to take forever
Truth: We will begin to see the impact of SDN in 2013.
It will take a number of years to see the full impact of SDN, but you can start right away. The first step, Centralizing Management, is something that can begin now. Networks today are frequently managed as individual devices; by installing a centralized management system, like Security Design – which is based on Juniper’s Junos Space management platform – you can begin to see the benefits of SDN right now.
There’s been lots of rhetoric and confusion around SDN. Now is the time to get past that confusion, to get clear on how SDN will transform business and begin to implement SDN within our networks. Juniper has a clear plan to deliver SDN to the industry. We’re the only company to clearly lay out the principles of SDN and the steps to get there. Juniper is also the only company to define a comprehensive, usage-based software licensing and pricing model that enables you to purchase the SDN services you need, when you need them.
We’ve released quite a bit of material about SDN this week. To learn more about the details of SDN, read my blog titled Decoding SDN.
To learn more about SDN Software Licensing, check out the Value Creation with SDN blog written by Brad Brooks, who runs marketing and business strategy for the Software Division at Juniper.
Or if you have some time, watch the Decoding SDN video which goes into a bit more detail.
Exploring the vision for the networking industry and the issues shaping its future.