 Now that we are well-versed in understanding how software-defined networks actually are configured If you consider a real-world deployment of an enterprise How would you consider migrating to software-defined networking? Would it be a gradual process? Would it be a one-time flag day event or would it be a mix of the two? Well, we have a usenix paper published back in 2014 by D11. It was published in a very prestigious conference usenix This paper talks about the approaches which have been invoked and If we adopt software-defined networking, how could we possibly migrate in a stepwise manner without giving away the production and the functionality of an operational legacy system So existing enterprise networks are expected to orchestrate better user-oriented services through software-defined networking Since most of the hardware is non SDN compliant and we need to incorporate SDN, we must talk about legacy systems and SDN-enabled networking equipment So considering these two and deploying them in order to Not give up on the functionality of legacy systems Challenge is the deployment because we are interested in making sure that the SDN implementation Is regressively robust For that Panopticon The the research paper proposes the combination of legacy and SDN switches How it realizes it is it provides a user interface or an abstraction of a logical software-defined networking view of the entire network which is partially upgraded legacy network Uh, this is not a perfect situation because we have legacy system as well So some resource constraints have to be taken into consideration imagine we we had Software-defined network with legacy systems. How could we possibly? realize in in addition to Panopticon we have a dual stack dual stack actually means that The legacy systems And software-defined networking are implemented as silos So there's no interaction between the two and this particular scheme is robust but This in the larger interest of an enterprise splits it into unwanted islands Then we have full edge SDN deployment in this case the end-to-end control is enabled through SDN while only using the traditional networking elements as as they are using their native operation This might seem plausible Because we are only considering the deployment on end devices This is not exactly how we could possibly think about network For that Panopticon is a partially deployed SDN which provides an interface to the network administrators and users to act more like a quasi SDN deployment So this is what we can look at it visually. We have dual stack where there's no overlap or traffic flow between SDN and legacy systems Then we have a SDN platform that provides legacy network infrastructure to behave kind of SDN oriented by working only on the edge And then we have Panopticon where we are having an immersed experience where we have the network elements comprising both legacy and SDN enabled hardware And then we have the SDN platform on the access devices which are all SDN enabled Let's now look at the possible steps that would be involved in its implementation First of all, there's an underlying assumption that SDN is not available on every switch So what is our goal? Our goal is to provide an interface to the users, which is SDN enabled All the traffic to and from the SDN connected ports would always be restricted to something known as a safe path Safe path actually means whatever we do the traffic must root through At least one SDN switch They call the authors call it a waypoint enforcement and they realize this through Splitting the entire network into a kind of Cellular arrangement or you can think about Voronoi diagram Where we have blocks each comprising at least one SDN element at the edge of every cell block Now the goal is to look at it as As an optimization problem where we are connecting the network elements of a certain graph By removing the SDN switches and links and making sure that These connected network components stay in In an overlaid form for the SDN In order to implement this we need a very smart network wide policy on the hardware the hardware is Assumed to be The layer two devices like ethernet switches would would be implementing layer two forwarding learning spanning tree protocols Vlan identifiers such as a to 2.1 q and per lan per vlan stp We have got routers Not all these routers and switches are SDN enabled. So what happens is we have a situation where Look at the scenario here. We've got a physical topology and we've got its logical equivalent the physical topology is Comprising SDN and non SDN enabled devices network elements here. You can see we've got a cell block A we've got cell block B. We've got cell block C D and E Here you can see that We've got four as cell blocks each cell block has certain ports which are SDN controlled ports or SDN enabled ports. So we can think about the physical cell blocks to be connected to At least one SDN controlled switch That is why on the right hand side We we we achieve the logical topology in which the all SDN controlled ports are connected to SDN switches via something known as a pseudo wires pseudo wires actually means it's it's a basically blend of The traffic comprising SDN and non SDN enabled parts The paper that I was referring to is panopticon reaping the benefits of incremental deployment in enterprise networks It's a very well cited paper. You might as well like to have a look