What is the difference between SDN network and traditional network?（1）

First, comparative analysis of MPLS and SDN networks from the perspective of network technology evolution

Since 1982, TCP/IP became the Internet's predecessor to the ARPANET standard, it has expanded rapidly with the development of the Internet. Basically, the data network has been unified, and everything over IP has been achieved. IP network as the underlying network that carries the Internet, Internet of Things, cloud computing, and various future infinite data services such as VR, AI, etc. Its flexibility, scalability, and support for services can require the network to complete its own evolution faster. .

From the overall development of the IP network, the basics of the network can be clearly divided into three phases. Of course, the networks of these three phases coexist in the real world.

Pure IP network (IP beats other network layer protocols and becomes the mainstream of data networks)

MPLS based network

SDN network

Pure IP network

Back a few decades ago, when the IP network was born, IP protocols and IP networks still had many competitors, such as IPX, Appletalk, and Netbios. These protocols, which are now listening to a long time, are still very popular at the time. For example, IPX is far more popular than IP in enterprise networks. The complexity of the network and network equipment of the year is largely reflected in supporting multiple network layer protocols, supporting multiple network speeds and interface types from E1 to OC-12, OC-48. IP and Ethernet are just a few of the many options.

APPANET chooses the TCP/IP combination, the ease of use of IP, the low cost of Ethernet, and the virality of the Internet, in just 20 years. The powerful combination of these three is like Intel + Microsoft, sweeping the data network. It can be considered that the Everything Over IP of the application layer is everywhere. By the year 2000, the voice network, video network and data network can be said to be Everything Over IP.

Pure IP networks contribute significantly to unified network layer protocols and simplified transmission media.

Pure IP network forwarding

In essence, it is hop-by-hop based on the destination IP address.

The way to accomplish this is through the IGP + BGP routing protocol.

Additional network control measures: PBR, ACL, IP-Based QoS

The network at this stage mainly implements the purpose of connectivity. The intra-AS and inter-domain routes are obtained through the IGP and BGP routing protocols. The router or switch that forms the network forms a forwarding table based on the routing table. When an IP packet arrives at the interface, the inbound chip extracts the IP packet header, searches the forwarding table based on the destination IP address, and finds the outgoing port.

For many years, most corporate networks, campus networks, and the public networks of quite a few carriers have adopted this approach. The advantage is simple and most devices support it. With the large-scale use of routing protocols and large-scale deployment of devices, as the overall system robustness, scalability and stability, as well as network construction costs have been greatly optimized. The corresponding network device chip system is also becoming more and more complete.

But from a service perspective, it seems to be lacking. The device can be implemented through PBR (Policy-based routing) when the service is required to communicate with the network. But this is not a universal service. For a small number of temporary, non-exclusive requirements, you can use the CLI to configure on some nodes, but no one will be crazy to go through the PBR to do routing on the whole network. Network engineers often refer to an implementation such as PBR called "Feature."

In those years, vendors have a lot of features. Used for gimmicks, used for bidding, to educate customers. Used to solve some practical problems not so perfect.

Feature has two implications:

It’s not a casual A-cat, it’s a ability.

This feature is conditional for both the network and the device.

Another phenomenon is that when some requirements are not met by Feature, or the performance bottleneck of the device, or the functional bottleneck, the network engineer has designed various appraisals, superimposing various large boxes and small boxes with special functions, most famous. The firewall is there, there are 4-7 layers of switching, WAN optimization and so on. After all, these devices solve the problem of "addressing based on destination address" on the processing of IP packets.

In summary, in the era of pure IP networks, the basic requirements are met by the routing protocol +feature+appliance.

Know more useful tips or study guide related to IT certification exam and Network technology from SPOTO now.