Differences between layer-3 vswitches and vrouters

Both layer-3 switches and routers work on the layer-3 network. packet forwarding (or switching) based on IP addresses is not very different in principle. These two terms tend to be unified, the exchange router and route switch refer to the same thing. We can consider a layer-3 switch as a multi-port router.

However, from the perspective of the historical development of the two, we can still find some differences:

Vro has a long history. In the middle of 1980s, CISCO launched the industry's first vro. Vro is launched to achieve remote interconnection and intercommunication between different network segments. The so-called "routing" refers to the query table based on the destination address (such as IP address) of the network data packets (such as IP packet, to determine and forward the packet to the next hop to the destination ". Because of the complexity of routing algorithms, early routers generally process and forward data packets based on the CPU using the "single hop Clock" command. In this way, the efficiency of packet forwarding is relatively low. However, we should not forget that at that time, the interconnection between remote and wide areas was the first, and the large enterprises (such as banks) that needed to use routers were large enterprises ). On the one hand, the WAN link speed carrying data is relatively low (less than 1 e1 = 2 Mbps, such as PSTN, ISDN, DDN, frame relay, etc ),
On the other hand, the applications at that time were mainly unix host computing, so the speed of network transmission is not a big problem.

To sum up, traditional routers have three features:

CPU-based one-step clock Processing Mechanism
Ability to process complex routing algorithms and protocols
It is mainly used for low-speed data links of wide area networks.
In 1990s, the rapid development of PCs led to the rapid rise and wide application of Ethernet and TCP/IP protocols. As the enterprise's internal network grew larger and larger, the Internet has become the largest wide area network, the enterprise's remote interconnection has been increasingly switched from self-built (rental) private networks to the Internet. These changes directly lead to the emergence of vswitches, upgrades of vrouters, and the convergence of vrouters and vswitches.

The earliest Ethernet switch came out around 95 years ago, its predecessor was a bridge. The bridge works on the second layer of the network. One of its functions is to isolate the network broadcast storm. We know that the establishment of communication links between hosts over Ethernet relies heavily on a broadcast package. Strictly speaking, a broadcast package is only responsible for establishing and maintaining communications without transmitting valid data, when the number of hosts on the network increases, the number of broadcast packets increases exponentially, occupying too much bandwidth and reducing network efficiency (the number of hosts in an Ethernet segment cannot exceed 30, when the limit is exceeded, the bridge should be used for isolation ). The bridge is like a person with a left arm and a right arm. It determines whether to exchange frames between the left and right sides based on the destination MAC address of the frame on the Ethernet. The broadcast package has no destination address (meaning to everyone), so it is partitioned. We can understand a vswitch as a multi-port bridge. The host or CIDR block dedicated bandwidth connected to each of its ports is independent of each other, improving network efficiency. Vswitches work on the second layer of the network. The algorithm is relatively simple, and some chip manufacturers will soon solidify it to produce the core ASIC chip of the vswitch and release the hardware line speed switch.

To sum up, traditional vswitches have three features:

Based on ASIC hardware, line speed switching
Working on the second layer of the network, mainly Ethernet
It is mainly used for fast data exchange in LAN
With the further expansion of the enterprise's internal network, the number of hosts has increased to hundreds, and the broadcast storm on the third layer of the network has become a problem (authoritative statement: the number of hosts in a TCP/IP CIDR block cannot exceed 300 ). Traditional switches work on the second layer of the network and cannot isolate layer-3 broadcasts. One solution is to isolate vlan on the vswitch and then converge the VLAN to the upper-level router. The router can forward routes between different CIDR blocks (layer-3 CIDR blocks. This is the original routing + switching model. A typical product is Cisco's catalyst 5500 route switch. Routing + switching is a transitional product. Today, layer-3 switches have greatly improved the architecture. Routing and switching are fully integrated through ASIC. After the first packet is determined by the routing algorithm, subsequent data packets are forwarded to the switching engine for line speed switching. Of course, it is still very difficult to solidify all routing algorithms. This is why the three-tier switch today cannot achieve line rate switching at 100%, or only supports limited and simple routing protocols.

Traditional routers continue to maintain their vitality with the development of the Internet. Complex protocols (such as BGP and MPLS), new features and applications (such as VPN and security) are inseparable from the computing of software and CPU, however, the router architecture has also been greatly improved and integrated with more and more firmware and ASIC.

The development of routing and exchange analyzes the evolution of software and hardware in the IT industry, and will continue to evolve with the relentless pursuit of functions and performance.