In recent years, the Internet has developed extremely rapidly, and applications are becoming increasingly commercialized, making the development of the number of online users difficult to predict. In addition, more and more users need high-speed access. Relevant information shows that in my country, slow Internet access is the primary problem that many netizens complain about. Therefore, improving network bandwidth, network service quality, and network management systems on routers are becoming increasingly important. On the premise of ensuring quality, using bandwidth to the maximum extent, detecting and diagnosing equipment failures early, and changing configurations and other network management functions quickly and conveniently as needed have become an important factor that directly affects the interests of network users and network operators. In general, the structure of routers is developing in three directions: faster speed, better service quality and easier integrated management.
Two major functions of router
Data path function: For each data packet arriving at the router, it is responsible for finding the way without losing it. This function mainly includes: forwarding decisions, output link queue scheduling via the backplane. The forwarding function is implemented through special hardware, and each packet packet passing through the router must perform this operation. Data path functionality is important for improving router performance.
Control functions: mainly include the management of routing tables and system configuration and management, as well as exchanging routing table information with adjacent routers, and implementing it through software. These features are not for each packet and are therefore used relatively less frequently.
Key technical issues of routers
Regarding some key technologies in router design at present, some problems have achieved phased results, and some have just begun research. It should be noted that this field is developing very fast. In order to facilitate systematic understanding of these key technical problems, we roughly divide these problems into three categories based on the three major trends in router development. First, issues related to high speed, such as fast routing table search, bus backplane and switching structure. The second is QoS problems, such as the classification of data flows and packets, traffic engineering and blocking control, etc. The third is software-related issues, such as software stability, configuration and management. It should be noted that this classification method is not strict, and some problems can belong to two categories at the same time.
1. Techniques related to speed
＊ Quick search technology for routing tables
In order to solve the shortage of address resources, reduce the scale of routing tables, and reduce management difficulty, the Internet adopts CIDR (Classless Inter-Doma in Routing). In this way, the routing table does not store specific IP addresses, but variable-length network prefixes. When addressing IP packets, the router adopts the longest network prefix matching (LPM-Longest Prefix Matching). For example, suppose there are two table entries "202., output port 1" and "202.168., output port 2" (X means any), if there is an IP packet with the destination address of 202.168.16.5, then this packet should be output from port 2. Traditional routers perform long-term network prefix matching, making routing table search a bottleneck in router speed. In the past two years, some quick table lookup algorithms have appeared, which can support Gigabit links. These algorithms include improved precise matching method, trie-based method, parallel hardware method, protocol-change simplified routing table query and buffering method, etc. Some of these algorithms are easy to implement in hardware, while others are suitable for software implementation. For multicast addressing, the table should be checked based on the source address and multicast address of the IP packet, the longest prefix matching method should be used for the source address, and the exact matching method should be used for the destination address.
＊ Exchange structure and scheduling algorithm
There are three ways to switch structures: Crossbar, shared memory and bus. The speed of the Crossbar structure is determined by the scheduler, the speed of the shared memory structure is determined by the read and write speed of the memory, and the speed of the shared bus structure is determined by the capacity of the bus and the overhead of arbitration. The scheduler is the core of the Crossbar switching structure. It collects information about the packet queue of each input port in each scheduling slot. Through a certain scheduling algorithm, a match between the input port and the output port is obtained, providing a path from the input port to the output port. Crossbar, which uses input buffering, and the use of ESLIP algorithm to implement scheduling has been adopted by some manufacturers. The difficulty of scheduler design is that it must not only meet the requirement of 100% system throughput, but also support CoS (Classes of Service). Among the scheduling algorithms, the weighted fair queuing algorithm (WFQ-Weighted Fair Queuing) and the improved weighted fair flow queuing algorithm (WF2Q-Weighted Fair Flow Queuing) are relatively easy to implement, and the performance is also good. In short, in order to enable the Internet to truly integrate multiple services, as a router connecting to core devices in the network, it is necessary to increase the port speed and switching capacity and provide QoS guarantee and traffic engineering (TE-Traffic Engineering) functions.
＊ Backplane bus
Due to the high port speed, the switching structure needs to be accelerated, and there are many connections between the interface board and the switching board, between the control board and the interface board, and between the control board and the switching board. Current high-speed routers and switches all use high-speed backplane buses to improve the system throughput. In order to improve efficiency, the data line uses a serial differential line, and the speed can reach 1 to 2 Gigabits per second. To ensure such a high transmission rate and so many connections, the design requirements for the backplane bus are very high. How to reduce the data transmission bit error rate and ensure clock synchronization are key technologies in bus design. In the Tiny Tera router project at Stanford University, a large number of implementation methods that use serial buses are worthy of research and reference.
2. Technology related to service quality
＊ Packet classification technology
The router must identify and classify the arrival of packet packets to determine the type of service it should accept. The solution IETF considered was at the core of the network, and the packets were identified based on the TOS (Type of Service) domain of the IP header. However, during the development of the Internet, since the terminal did not consider TOS when sending IP packets, TOS has never played a role. Currently, on edge devices, packets are identified based on the source IP address, destination IP address, source port number, destination port number, and transport layer protocol type of the IP packet. In addition, in order to realize the function of the firewall, IP packets are also required.
During identification, each identification rule adopts the source IP address, destination IP address, source port number, destination port number, and transport layer protocol type. In the above identification rules, each domain may be an interval. For example, there is an identification rule "202.66., 202.66., X, 23, TCP" (X means any), which identifies telnet data from network 202.66. to network 202.66. From a geometric perspective, if the K domains of the IP header are used during discrimination, this problem is that there are actually many overlapping entities in a K-dimensional space (each discrimination rule corresponds to an entity). Whenever a packet arrives, the packet is equivalent to a point in the K-dimensional space. Discrimination is actually to find the entity with the highest priority that contains the point.
＊ Classification of data flows
A series of packets through a given source and destination are used as data streams, which can be a series of packets of TCP connections maintained for a long time, or a series of UDP packets of sound or image. Generally, there are two standards for streams: port pairs and host pairs. By port, the data packets of the same stream must have the same source, destination address, TCP/UDP port number, etc.; by host, the data packets must only have the same source and destination address. There are three commonly used stream classifiers at present, one is the X/Y classifier, Y is the specified time interval and X is the number of packets. If the number of packets arrived by a certain data stream in time Y is greater than X, the stream is considered a long data stream, otherwise it is a short data stream; the other is a protocol classifier, which stipulates that all TCP packets are defined as data streams; the other is a port classifier, which specifies several special TCP ports as long data streams. Because classification requires header checking for each package entering the router, we need a fast classification algorithm.
* QoS issues
If true QoS can be achieved, it is of course the most ideal for the Internet to become a commercial application network. In the current situation where network blockage is inevitable, the IETF has proposed some models and mechanisms to support QoS, with two main implementation frameworks, namely IS (Integrated Service) and DS (Differentiated Service). Among them, IS applies the resource reservation protocol (RSVP), establishes a sending channel and reserves resources before real-time service transmission; while DS classifies data packets by marking data packets, so that different data packets can enjoy different services. It is difficult to implement true QoS. Whether it is the switching structure that uses output buffering or shared buffering, N times acceleration is required internally.
＊MPLS and traffic engineering
On the Internet, explicit routing can be supported via MPLS. Explicit routing can make up for the lack of the OSPF (Open Shortest Path First) protocol in traditional IP networks that only establishes one shortest path between the source and the destination. Its advantage is that ISP can allocate traffic on different paths between two points, thereby reducing the blocking rate of busy paths, improving the utilization rate of lines, and providing a certain degree of QoS guarantee. Redundant lines can also be established through explicit routing to ensure smooth communication lines in the event of some router failure. The relevant working groups of the IETF and many network equipment manufacturers are doing this and have proposed some implementation solutions.
＊ Blocking control
The TCP protocol uses a sliding window mechanism for traffic and blocking control, and in fact, the sending end adjusts its own transmission rate according to the network blocking situation.
The TCP protocol has two characteristics: one is unfairness. The short process sending window of RTT (Round Trip Time) increases rapidly and will occupy more bandwidth. Routers should suppress this injustice. Another is the synchronization phenomenon. When the tail drop strategy adopted by the previous routers blocked the network, it may cause the sending windows of many TCP connections to drop sharply at the same time and then rise at the same time, resulting in blockage; in this cycle, the utilization rate of the line is very low. This phenomenon is called synchronization, and the router should take measures to avoid this phenomenon.
One of the main measures for routers to handle blocking is to utilize the packet loss nature of the TCP protocol. When packets are lost, TCP reduces the sending rate, thereby reducing the burden on the router. Since the TCP protocol is very sensitive to packet loss, packet loss will cause a sharp drop in the transmission rate. The router should optimize buffer management for TCP performance to avoid a sharp decline in network performance.
Currently, many routers use the RED Random Early Detection method to perform blocking control. This method reduces the transmission rate by dropping packets early before blocking occurs, so as to reduce network load, avoid blocking, and suppress synchronization. The biggest disadvantage of the RED algorithm is that it only works on the transport layer protocol that automatically controls flow like the TCP protocol; packet loss for UDP protocol routers will not prompt the sender to reduce the transmission rate. Since TCP data traffic accounts for a large proportion of the network, the performance of the RED algorithm is relatively good. Many routers on the Internet use this method for blocking control, and this method has proven to be effective.
3. Software-related issues
＊The importance of software
In the past, routers were regarded as the best hardware device for forwarding packets, and the software only provides the functionality of a monitor. But with the development of routers, software plays an increasingly important role in routers. In fact, the choice of real-time operating systems (such as PSOS and VxWorks commonly used in the communications field) is crucial for a communications product. If you want to develop highly efficient software, you need the support of the operating system manufacturer. For example, Cisco is developing its own dedicated router operating system and application software. If this trend continues to develop, end users can easily load various application software modules on the router in the future, so that the router can provide functions such as firewall, traffic management policies, special application signaling, routing policies, etc.
＊Network Management System
The current network management protocol is Simple Network Management Protocol version 2.0 (SNMPv2, Simple Network management Protocol v2.0), which is implemented using UDP protocol in the TCP/IP protocol. Due to the changes in the architecture of the router, some network management information needs to be provided by the underlying hardware, which is different from the previous implementation method. The work here mainly provides network managers with sufficient management information and powerful and flexible management functions through the implementation of network management functions and management information base (MIB-Management Information Base).
＊ Billing
To bill the user's data traffic, you need to extract the IP packet's address, port, CoS and other information. Because the port speed is very high, the amount of data in this part of the information is also relatively large, and it is unrealistic to process it on the interface board. Moreover, the charging standards for different ISPs are likely to be different. Therefore, the billing function should be separated, and the router interface board is only responsible for providing an interface for billing information and sending the billing information out. After that, the data can be written to an external storage device and billed by a special machine. This will reduce the burden on the router and implement the billing function more flexible.
＊Configuration
The configuration of a router is a very important and difficult task. Once an incorrect configuration occurs, it is not only difficult to detect, but also some difficult performance problems will occur. With the development of router technology, its configuration will become simpler and more effective, and the perfect solution to this problem will be a long-term task.
＊Software
As we all know, the router's hardware can improve stability by using hot backup, dual power supply, data path backup, etc., but the stability of the software is a difficult problem to solve. The prerequisite for the stability of a large network system is the stability of the software. The difficulty of software stability is that the state of the software is affected by the interactions between different software.
Conclusion
From a global perspective, based on the current network development, Gibit routers are the mainstream of the high-speed router market, and Taibit routers are not large. But judging from the development momentum of the Internet, Taibit routers will be useful, and some companies are already applying and testing it. For example, the US GST company has decided to use Avici's terabyte router on its "SuperNet" network.
A research report released by Internet measurement and statistics company I Amasia in * on June 7, 2000 pointed out that the number of Internet users in my country has reached 12.3 million. In mainland my country, the number of users who use the Internet at home is currently 5.7 million, while the number of people who use the Internet at home is 5 million; while the number of people who use the Internet at home and in the office is 1.52 million and 610,000 respectively. my country's network scale is not large now. Recently, the development of backbone networks uses Gibit routers, and 163 network uses 6 Cisco Gibit routers. But in a great country like us, the development potential of users will be immeasurable in the next few years. Some backbone nodes need to consider tybit routers, and gigabit routers may be used on secondary nodes or edge nodes.
At present, there is an urgent need for routers that can provide classified business functions. The activation of IP phones shows this. If the router can implement this function, not only will the quality of IP phones be much better than now, but the Internet will also truly become a multimedia network, thus bringing more commercial value. I believe that all this will become a reality in the near future.
The companies that have been developing router products in my country are Huawei and Sunda. Recently, some powerful telecommunications companies and scientific research institutions in China have just entered the field of high-speed and high-performance next-generation router research. my country has had successful experience in the field of ATM switches. Scientific researchers have developed products similar to international standards in a short period of time and have been put into commercial use. In terms of the research and development of next-generation routers, my country has the hope of catching up with the world's advanced level. I believe that our country's next-generation router products can occupy a place in the market as soon as possible.
Two major functions of router
Data path function: For each data packet arriving at the router, it is responsible for finding the way without losing it. This function mainly includes: forwarding decisions, output link queue scheduling via the backplane. The forwarding function is implemented through special hardware, and each packet packet passing through the router must perform this operation. Data path functionality is important for improving router performance.
Control functions: mainly include the management of routing tables and system configuration and management, as well as exchanging routing table information with adjacent routers, and implementing it through software. These features are not for each packet and are therefore used relatively less frequently.
Key technical issues of routers
Regarding some key technologies in router design at present, some problems have achieved phased results, and some have just begun research. It should be noted that this field is developing very fast. In order to facilitate systematic understanding of these key technical problems, we roughly divide these problems into three categories based on the three major trends in router development. First, issues related to high speed, such as fast routing table search, bus backplane and switching structure. The second is QoS problems, such as the classification of data flows and packets, traffic engineering and blocking control, etc. The third is software-related issues, such as software stability, configuration and management. It should be noted that this classification method is not strict, and some problems can belong to two categories at the same time.
1. Techniques related to speed
＊ Quick search technology for routing tables
In order to solve the shortage of address resources, reduce the scale of routing tables, and reduce management difficulty, the Internet adopts CIDR (Classless Inter-Doma in Routing). In this way, the routing table does not store specific IP addresses, but variable-length network prefixes. When addressing IP packets, the router adopts the longest network prefix matching (LPM-Longest Prefix Matching). For example, suppose there are two table entries "202., output port 1" and "202.168., output port 2" (X means any), if there is an IP packet with the destination address of 202.168.16.5, then this packet should be output from port 2. Traditional routers perform long-term network prefix matching, making routing table search a bottleneck in router speed. In the past two years, some quick table lookup algorithms have appeared, which can support Gigabit links. These algorithms include improved precise matching method, trie-based method, parallel hardware method, protocol-change simplified routing table query and buffering method, etc. Some of these algorithms are easy to implement in hardware, while others are suitable for software implementation. For multicast addressing, the table should be checked based on the source address and multicast address of the IP packet, the longest prefix matching method should be used for the source address, and the exact matching method should be used for the destination address.
＊ Exchange structure and scheduling algorithm
There are three ways to switch structures: Crossbar, shared memory and bus. The speed of the Crossbar structure is determined by the scheduler, the speed of the shared memory structure is determined by the read and write speed of the memory, and the speed of the shared bus structure is determined by the capacity of the bus and the overhead of arbitration. The scheduler is the core of the Crossbar switching structure. It collects information about the packet queue of each input port in each scheduling slot. Through a certain scheduling algorithm, a match between the input port and the output port is obtained, providing a path from the input port to the output port. Crossbar, which uses input buffering, and the use of ESLIP algorithm to implement scheduling has been adopted by some manufacturers. The difficulty of scheduler design is that it must not only meet the requirement of 100% system throughput, but also support CoS (Classes of Service). Among the scheduling algorithms, the weighted fair queuing algorithm (WFQ-Weighted Fair Queuing) and the improved weighted fair flow queuing algorithm (WF2Q-Weighted Fair Flow Queuing) are relatively easy to implement, and the performance is also good. In short, in order to enable the Internet to truly integrate multiple services, as a router connecting to core devices in the network, it is necessary to increase the port speed and switching capacity and provide QoS guarantee and traffic engineering (TE-Traffic Engineering) functions.
＊ Backplane bus
Due to the high port speed, the switching structure needs to be accelerated, and there are many connections between the interface board and the switching board, between the control board and the interface board, and between the control board and the switching board. Current high-speed routers and switches all use high-speed backplane buses to improve the system throughput. In order to improve efficiency, the data line uses a serial differential line, and the speed can reach 1 to 2 Gigabits per second. To ensure such a high transmission rate and so many connections, the design requirements for the backplane bus are very high. How to reduce the data transmission bit error rate and ensure clock synchronization are key technologies in bus design. In the Tiny Tera router project at Stanford University, a large number of implementation methods that use serial buses are worthy of research and reference.
2. Technology related to service quality
＊ Packet classification technology
The router must identify and classify the arrival of packet packets to determine the type of service it should accept. The solution IETF considered was at the core of the network, and the packets were identified based on the TOS (Type of Service) domain of the IP header. However, during the development of the Internet, since the terminal did not consider TOS when sending IP packets, TOS has never played a role. Currently, on edge devices, packets are identified based on the source IP address, destination IP address, source port number, destination port number, and transport layer protocol type of the IP packet. In addition, in order to realize the function of the firewall, IP packets are also required.
During identification, each identification rule adopts the source IP address, destination IP address, source port number, destination port number, and transport layer protocol type. In the above identification rules, each domain may be an interval. For example, there is an identification rule "202.66., 202.66., X, 23, TCP" (X means any), which identifies telnet data from network 202.66. to network 202.66. From a geometric perspective, if the K domains of the IP header are used during discrimination, this problem is that there are actually many overlapping entities in a K-dimensional space (each discrimination rule corresponds to an entity). Whenever a packet arrives, the packet is equivalent to a point in the K-dimensional space. Discrimination is actually to find the entity with the highest priority that contains the point.
＊ Classification of data flows
A series of packets through a given source and destination are used as data streams, which can be a series of packets of TCP connections maintained for a long time, or a series of UDP packets of sound or image. Generally, there are two standards for streams: port pairs and host pairs. By port, the data packets of the same stream must have the same source, destination address, TCP/UDP port number, etc.; by host, the data packets must only have the same source and destination address. There are three commonly used stream classifiers at present, one is the X/Y classifier, Y is the specified time interval and X is the number of packets. If the number of packets arrived by a certain data stream in time Y is greater than X, the stream is considered a long data stream, otherwise it is a short data stream; the other is a protocol classifier, which stipulates that all TCP packets are defined as data streams; the other is a port classifier, which specifies several special TCP ports as long data streams. Because classification requires header checking for each package entering the router, we need a fast classification algorithm.
* QoS issues
If true QoS can be achieved, it is of course the most ideal for the Internet to become a commercial application network. In the current situation where network blockage is inevitable, the IETF has proposed some models and mechanisms to support QoS, with two main implementation frameworks, namely IS (Integrated Service) and DS (Differentiated Service). Among them, IS applies the resource reservation protocol (RSVP), establishes a sending channel and reserves resources before real-time service transmission; while DS classifies data packets by marking data packets, so that different data packets can enjoy different services. It is difficult to implement true QoS. Whether it is the switching structure that uses output buffering or shared buffering, N times acceleration is required internally.
＊MPLS and traffic engineering
On the Internet, explicit routing can be supported via MPLS. Explicit routing can make up for the lack of the OSPF (Open Shortest Path First) protocol in traditional IP networks that only establishes one shortest path between the source and the destination. Its advantage is that ISP can allocate traffic on different paths between two points, thereby reducing the blocking rate of busy paths, improving the utilization rate of lines, and providing a certain degree of QoS guarantee. Redundant lines can also be established through explicit routing to ensure smooth communication lines in the event of some router failure. The relevant working groups of the IETF and many network equipment manufacturers are doing this and have proposed some implementation solutions.
＊ Blocking control
The TCP protocol uses a sliding window mechanism for traffic and blocking control, and in fact, the sending end adjusts its own transmission rate according to the network blocking situation.
The TCP protocol has two characteristics: one is unfairness. The short process sending window of RTT (Round Trip Time) increases rapidly and will occupy more bandwidth. Routers should suppress this injustice. Another is the synchronization phenomenon. When the tail drop strategy adopted by the previous routers blocked the network, it may cause the sending windows of many TCP connections to drop sharply at the same time and then rise at the same time, resulting in blockage; in this cycle, the utilization rate of the line is very low. This phenomenon is called synchronization, and the router should take measures to avoid this phenomenon.
One of the main measures for routers to handle blocking is to utilize the packet loss nature of the TCP protocol. When packets are lost, TCP reduces the sending rate, thereby reducing the burden on the router. Since the TCP protocol is very sensitive to packet loss, packet loss will cause a sharp drop in the transmission rate. The router should optimize buffer management for TCP performance to avoid a sharp decline in network performance.
Currently, many routers use the RED Random Early Detection method to perform blocking control. This method reduces the transmission rate by dropping packets early before blocking occurs, so as to reduce network load, avoid blocking, and suppress synchronization. The biggest disadvantage of the RED algorithm is that it only works on the transport layer protocol that automatically controls flow like the TCP protocol; packet loss for UDP protocol routers will not prompt the sender to reduce the transmission rate. Since TCP data traffic accounts for a large proportion of the network, the performance of the RED algorithm is relatively good. Many routers on the Internet use this method for blocking control, and this method has proven to be effective.
3. Software-related issues
＊The importance of software
In the past, routers were regarded as the best hardware device for forwarding packets, and the software only provides the functionality of a monitor. But with the development of routers, software plays an increasingly important role in routers. In fact, the choice of real-time operating systems (such as PSOS and VxWorks commonly used in the communications field) is crucial for a communications product. If you want to develop highly efficient software, you need the support of the operating system manufacturer. For example, Cisco is developing its own dedicated router operating system and application software. If this trend continues to develop, end users can easily load various application software modules on the router in the future, so that the router can provide functions such as firewall, traffic management policies, special application signaling, routing policies, etc.
＊Network Management System
The current network management protocol is Simple Network Management Protocol version 2.0 (SNMPv2, Simple Network management Protocol v2.0), which is implemented using UDP protocol in the TCP/IP protocol. Due to the changes in the architecture of the router, some network management information needs to be provided by the underlying hardware, which is different from the previous implementation method. The work here mainly provides network managers with sufficient management information and powerful and flexible management functions through the implementation of network management functions and management information base (MIB-Management Information Base).
＊ Billing
To bill the user's data traffic, you need to extract the IP packet's address, port, CoS and other information. Because the port speed is very high, the amount of data in this part of the information is also relatively large, and it is unrealistic to process it on the interface board. Moreover, the charging standards for different ISPs are likely to be different. Therefore, the billing function should be separated, and the router interface board is only responsible for providing an interface for billing information and sending the billing information out. After that, the data can be written to an external storage device and billed by a special machine. This will reduce the burden on the router and implement the billing function more flexible.
＊Configuration
The configuration of a router is a very important and difficult task. Once an incorrect configuration occurs, it is not only difficult to detect, but also some difficult performance problems will occur. With the development of router technology, its configuration will become simpler and more effective, and the perfect solution to this problem will be a long-term task.
＊Software
As we all know, the router's hardware can improve stability by using hot backup, dual power supply, data path backup, etc., but the stability of the software is a difficult problem to solve. The prerequisite for the stability of a large network system is the stability of the software. The difficulty of software stability is that the state of the software is affected by the interactions between different software.
Conclusion
From a global perspective, based on the current network development, Gibit routers are the mainstream of the high-speed router market, and Taibit routers are not large. But judging from the development momentum of the Internet, Taibit routers will be useful, and some companies are already applying and testing it. For example, the US GST company has decided to use Avici's terabyte router on its "SuperNet" network.
A research report released by Internet measurement and statistics company I Amasia in * on June 7, 2000 pointed out that the number of Internet users in my country has reached 12.3 million. In mainland my country, the number of users who use the Internet at home is currently 5.7 million, while the number of people who use the Internet at home is 5 million; while the number of people who use the Internet at home and in the office is 1.52 million and 610,000 respectively. my country's network scale is not large now. Recently, the development of backbone networks uses Gibit routers, and 163 network uses 6 Cisco Gibit routers. But in a great country like us, the development potential of users will be immeasurable in the next few years. Some backbone nodes need to consider tybit routers, and gigabit routers may be used on secondary nodes or edge nodes.
At present, there is an urgent need for routers that can provide classified business functions. The activation of IP phones shows this. If the router can implement this function, not only will the quality of IP phones be much better than now, but the Internet will also truly become a multimedia network, thus bringing more commercial value. I believe that all this will become a reality in the near future.
The companies that have been developing router products in my country are Huawei and Sunda. Recently, some powerful telecommunications companies and scientific research institutions in China have just entered the field of high-speed and high-performance next-generation router research. my country has had successful experience in the field of ATM switches. Scientific researchers have developed products similar to international standards in a short period of time and have been put into commercial use. In terms of the research and development of next-generation routers, my country has the hope of catching up with the world's advanced level. I believe that our country's next-generation router products can occupy a place in the market as soon as possible.
[1]
Article entry: csh Editor in charge: csh