TRADITIONAL COMMUNICATIONS
In legacy, the communication between two devices had to be sent on various transport medium like SONET, Mux, Frame-Relay, Phone-lines however they are very high cost also multi-user transport was not well managed in turn causing high resource utilization and complicated control plane management for end users. Even the end devices management of large scale requires time consuming task on each devices and hoigh resource management as well for operations. High Capex and Opex.
MPLS COMMUNICATIONS
In MPLS, the transport domain is well organized to carry multi-protocol datagrams of multiple customer in distinguished way to identify source & destination and bring reachability. The far distance communication like inter-continent or inter-country are well maintained with efficient control plane centrally managed by Route-Reflector clusters with many redundancy. The end customer control plane information’s are carried on a MPLS transport routing protocol called MP-BGP by distinguishing themselves with Route-Target and Route-Distinguisher. However this has been a costlier for customers and had too rely on this transport domain for their VPN communications. High Opex.
The end user has to rely on MPLS domain for their VPN and control plane will have to make the calculations based on number of links they have with MPLS Provider as next-hop. There wasn’t a centralized operations or automation been carried in this time period, the resource usage at end site was shared and efficient and it simplified multi-user communication bit easier with common resources. But to make use of Internet Application along with VPN was always been tough one to handle in this setup. A site typical would have a Firewalls, Many Routers for VPN and Internet Traffic with Redundancy, Distributed Switches, WAN optimizer and so on.
SDWAN COMMUNICATIONS
Long distance VPN communications had been made possible irrespective of all transport domain with central management in SDWAN. To Reach R4 the control plane will have next-hop R4 system IP/Router IP directly that is a common single TLOC System IP irrespective of number of transport interfaces in the router. Now the transport domain responsibility is to bring reachability and has a IPSEC Tunnel kept ready to reach all Devices IP in SDWAN Fabric. Now WAN is efficiently managed with destination router reachability by TLOC System IP irrespective of the Transport. This gives efficient usage of transport domain with high redundancy and end vEdges are well managed by centralized location with Controllers performing many operations for vEDGE functioning like Config Management, Performance Check, Utilization Check, Control Plane Management etc all kind of operations in single pain of Glass for user. However there are certain limitations in Multi-Customer data transport and communications of devices in large scale in Singe Management of operations.
If you see the network diagram below no matter how many complicated transport involved the end user will always have a tunnel kept ready with end devices and communications happens with immediate switchover and efficient load-balancing also with central management of these end devices. VPN is having high redundancy with maximum use of transport and Internet communications to cloud applications are also improvised with traffic prioritization, all these are done through singe pain of glass management. End customer have the control to modify the traffic flow with GUI in eased manner.
There are four Planes in SDWAN which helps in managing the end devices:-
1. Management Plane
Management Plane consist of vManage, which is always Virtual deployed on a server. one vManaage can handle 2000 WAN Edges, if it has to be scaled more then can use 1 Cluster of three vManage so this make it capable of 6000 Edge devices handling. All Data processing and Config management decision are taken through vManage Portal and instructions will be sent to vSmart to act on towards to any Edge devices. Management plane is merely a user presentation for handling all the SDWAN Fabric and make selection on the requirement.
2. Control Plane
vSmart is the brain of SDWAN Fabric, which take cares of control plane operation. vSmart will be Virtual and can manage upto 5400 control connections that is the transport connection towards edge devices. We can form 20 vSMART cluster if needed. vSmart uses the route exchange protocol called OMP overlay management protocol to fetch VPN routes and give it to other vEdges. Like for example vEdge1 has 5 LAN routes then it will be given to vSmart via the DTLS overlay tunnel with the help of OMP Protocol and in turn vSmart share those 5 routes to all the vEdges in the fabric with next-hop WAN IP of each transport in vEdge1. If vEdge has the IPsec tunnel up via all the transport to reach next-hop then those 5 routes will be reachable to all vEdges and can do load balance however we also can manipulate the traffic as we want.
Even if the VSmart goes down also the these IPSec tunnel will remain UP and send traffic as long as transport WAN IP is reachable to all vEdges.
3. Orchestration Plane
Orchestration plane consist of vBond, this component acts in the initial authentication and on-boarding of edge devices. When a edge device is first plugged-in vBond is the only devices that it knows to communicate to form connection/Validation phase. vBond can be physical or Virtual, it uses the image of vEdge i.e. cEdge/vEdge. Please note before it can form a authentication process with vBond, it has to be certified by an authority. vBond verifies the certification and validates the serial and chassis with its own database and gives an approval that device is valid and been sent control details to vEdges and also send vEdge details to Controllers vice versa. This on-boarding needs high level of database administration hence kept in separate plane called Orchestration plane.
4. Data Plane
Data plane are more about edge devices having control plane build up to form the forward plane towards any transport that it is connected. IPSec tunnels build upon all transport as a overlay connection with each edge devices in the fabric having associated with that VPN is the Forward Plane ready for Data transmission.
End devices in Cisco Viptela can be Cisco ISR 1K,4K & 8K. ASR 1K or ENCS 5000. Viptela 100,1000, 2000, 5000. Virtual devices can be CSR 1000v,Cisco ISRv, Vedge Cloud. Versa it can Advantech devices and so ON with different vendors.
Now we have the control information exchanged to all vEdges with the help of those planes explained above, if I put in simple terms as long as transport WAN IP’s are reachable with all vEdge devices then load balancing or traffic manipulation are done easy with the help of controllers. Edges are manipulated to our needs based on requirement of traffic sensitivity by the central management vManage.
Some Examples on Advantages of SDWAN below:-
Advantage 1:- If there are multiple customer terminated on single router like in below airport, then there would be multiple eBGP session to be formed for each customer, however In SDWAN WAN side would have single routing protocol just to give reach-ability to end devices. All the customer VRF routes are given to SDWAN controller by single connection. Now Controller does the distribution of each customer routes to the respective destined devices.
Advantage 2: PE routers in MPLS would not have customer routes, which saves alot of memory for PE also the control processing cycle. Imagine how much routes of each customer PE would have handle, now all the routes won’t be present. PE Routers will just handle the WAN Routes for each device to bring reach-ability, all the customer routes are now handled by SDWAN controllers and given to end SDWAN devices with a secured connection.
Advantage 3:- End site doesn’t need to have Firewall, WAN Optimizer, Load Balancer. All such features are covered in single edge devices which handles packet flow well optimized for customer.
Advantage 4:- Imagine there are 100 devices at the customer site, to configure remotely is time consuming and quality compromised task. However with SDWAN the configuration can be done through SDWAN GUI by updating the device specific values and complete configs are pushed in one go without any errors.
Advantage 5:- If there are multiple transport domain available for SDWAN devices at the customer site, then SDWAN has the capability to load balance the traffic on all links and bring TCP segmented packet flow in correct order even with irregular bandwidth. Traditional routers did not had that capabilities of re-arranging packets from multiple transport domain.
Advantage 6:- All the features that we see in end devices like the utilisation of link, Errors, CPU consumption, Power and Netflow etc.. can be done through single GUI administration, which graphically shows all in a simple way and gives a ability to configure too for each device. we call it as single pane of glass management for all devices.
Advantage 7:- Zero touch provisioning helps the end devices to plug and use instantly, all the device specific values or configuration are pre-created in Controllers hence if the values are verified then controller validates the end devices and push the configuration instantly. If values/Configs are pre-verfied correctly then device installation would be quicker eliminating the time taken for configuration or troubleshooting if done manually. Installation are been faster then traditional Router deployment at site.
Advantage 8:-
(4 votes, average: 5.00 out of 5)
Very good work prabhu, keep it up