Each parameter affects the performance, functionality, and different integration characteristics of the switch. 1. Forwarding Technology: The switch adopts direct forwarding technology or storage forwarding technology? 2, delay: How much is the switch data exchange delay? 3, management function: How much can the switch supplied to support? 4, single / multi-MAC address type: Each port is a single MAC address, or a MAC address? 5, external monitoring support: Does the switch allow external monitoring tools to manage port, circuits or switches all traffic? 6, extended tree: Does the switch provide extended tree algorithm or other algorithm, detect and limit the topological ring? 7, full duplex: Does the switch allow port to receive / send, full duplex communication? 8, high-speed port integration: Does the switch provide high-speed port connection key business servers or upside down? The following items are discussed item by item: 1) Forward Technologies) Forwarding Technology refers to the forwarding mechanism used to determine how to forward packets. Various forwarding technologies have advantages and disadvantages. Direct Forwarding Technology: (CUT-THROUGH) Switch Once interpreted to the packet destination address, it will start to send a packet to the destination port. Typically, the switch already knows the destination address while receiving the first 6 bytes of the packet, so that it can determine which port is forwarded to this packet. The advantage of direct forwarding technology is that the forwarding rate is fast, reducing delay and improves overall throughput. Its disadvantage is that the switch has begun data forwarding before the correctness is not fully received and checked the packet. In this way, in an environment with high communication quality, the switch forwards all full packets and error packets, which actually brings many garbage communication packages to the entire switch network, and the exchange opportunity is misunderstood to have a broadcast storm. In short, the connection between direct forwarding technologies is good, and there is a network environment with less network links and less packets. Storage Forwarding Technology: (Store-and-Forward) Storage Forwarding Technical Requirements The switch will decide how to forward it after receiving all packets. In this way, the switch can check the integrity and correctness before forwarding. The advantage is that there is no incomplete packet forward, which reduces potential unnecessary data forwarding. The disadvantage is that the forwarding rate is slower than the direct forwarding technology. Therefore, storage forwarding techniques are compared to network environments with normal link quality. Collision Escape Forwarding Technology: (Collision-Avoidance) Some vendors (3COM) switches also offer specific forwarding technologies for this vendor. Collision escape forwarding techniques have chosen a compromise between the high-transgent rate and the high correctness rate by reducing network errors. 2) Time: (Latency) switch delay means the time interval between the data packet to start copying the packet to the destination port from the switch. Many factors can affect delay size, such as forwarding technologies, etc. The switch with direct forwarding technology has a fixed delay. Because the pass-through switch regards only the integral size of the data packet, the forwarding direction is determined according to the destination address. Therefore, its delay is fixed, depending on the interpretation rate of the destination address in the first 6 bytes of the data packet. Switches that adopt storage forwarding technologies must begin forwarding the packet since the complete packet must be received, so its delay is related to the packet size. The packet is large, then the delay is large; the data package is small, the delay is small. 3) Management Function: The management function of the switch means how the switch controls the user access switch, and how the user's visual extent is visible. Typically, switch vendors provide management software or meet third-party management software remote management switches. The general switch meets the SNMP MIB I / MIB II statistical management function.
Combine exchange opportunities increase the RMON active monitoring function by built-in RMON group (Mini-RMON). Some switches also allow external RMON to detect network conditions of the optional port. 4) Single / multi-MAC address type: (Single- Versus Multi-Mac) Single MAC switch has only one MAC hardware address. Each port of the multimod switch is bundled with multiple MAC hardware addresses. Single MAC switches are mainly designed to connect end users, network sharing resources or non-bridge routers. They cannot be used to connect a hub or a network segment containing multiple network devices. Multi-MAC switches have sufficient storage to memory multiple hardware addresses in each port. Each port of the multimanic switch can be seen as a hub, while the MAC switch can be seen as a hub hub. The capacity size of the banks of each vendor's switch is different. The size of this Buffer capacity limits the exchange address capacity that this switch can be provided. Once this address capacity is exceeded, some switches will discard other address packets, and some switches copy the packet to each port does not exchange. 5) External Monitor Support: (Extendal Monitoring) Some switches provide "Monitoring Port", allowing an external network analyzer to directly connect to the switch to monitor network conditions. However, the implementation methods of each manufacturer are different. 6) Extended Tree: (Spanning Tree) Since the switch is actually a transparent bridge device of the multi-port, the switch also has the inherent problem of bridged devices - Topology Loops. When a data packet of a network segment is transmitted to another by a bridge device, the returned packet returns the source address through another bridge device. This phenomenon is called "Topology". Generally, the switch uses the expansion tree protocol algorithm to let each of the bridges in the network, automatically prevent the topology ring. The switch is disconnected by disconnecting a certain port in the detected "Topological ring" to achieve the purpose of the "Topological ring", and maintain the integrity of the topology tree in the network. In web design, "Topology Ring" is often recommended for redundant backup link selection for critical data links. So, switches with extended tree protocols can be used to connect to switching redundancy in key resources in the network. 7) Full-duplex: (Full Duplex) full-duplex ports can simultaneously send and receive data, but this switch and the connected device support full-duplex mode of operation. Switches with full-duplex function have the following advantages: 1. THROUGHPUT: twice the communication throughput of single work mode. 2. Avoid collision (Collision Avoidance): No transmission / reception collision. 3, breakthrough length limit (IMPROVED DISTANCE LIMITATION): Since there is no collision, it is not limited by the length of the CSMA / CD link. The length limit of the communication link is only related to the physical media. The protocols that now support the full-duplex communication include Quick Ethernet, Gigabit Ethernet and ATM. 8) High-speed port integration: (HIGH-SPEED Intergration) The switch can provide high bandwidth "pipe" (fixed port, optional module or multi-link tunnel) to meet the switching traffic of the switch and the exchange needs of the superiors. Prevent a backbone communication bottleneck. Common high-speed ports are: FDDI: Applications are earlier and wide. But there is a protocol conversion cost. FAS
T Ethernet / Gigabit Ethernet: Easy to connect, with a small amount of protocol conversion; but is limited by the network size. ATM: High-speed exchange ports are available; however, the protocol conversion cost is large. Second, ATM Switch (ATM Switch) With the development of ATM exchange technology, it is now increasingly using ATM exchange technology in the high-speed network backbone or edge network. According to the development requirements of existing enterprises, adapt to the technical trend of data network exchange, we need to know ATM. ATM data exchange consists of a fixed length ATM cell. Each ATM cell is 53-byte length (5 word-wide letterhead and 48-byte long confidence). The letterhead includes address information such as virtual path (VP) and virtual circuit (VC) identity. ATM determines the transmission source address of the cell and the receiving destination address according to the VP and VC. The connection in the ATM switch is divided into two types of permanent virtual circuit (PVC) and switch virtual circuit (SVC). The PVC is a permanent hardware circuit connection between the source address and the destination address. The SVC is a temporary exchange circuit established according to real-time exchange requirements. The greatest difference between the two is that the PVC remains a connection regardless of whether there is data transmission; the SVC is automatically disconnected after the data transfer is complete. The application difference between the two is: In the usual ATM exchange, some PVC is used to keep the signal and management information, keep a permanent connection; the SVC is mainly used for a large amount of specific data transmission. Another feature of ATM exchange is: ATM itself is a full-duplex. Sending data and receiving data simultaneously in different virtual circuits, maintaining a two-way high-speed communication. In order to meet the mutual communication requirements of Ethernet frames and ATM cells (Cells), the ATM protocol standard specifies the ATM Adaption Layer for data applications, which works between frame exchange and cell exchange. The address information of the logic circuit layer of the Ethernet frame corresponds to the virtual circuit VC, the virtual path VP address information, completed frame-cell conversion, and cell-frame conversion operations. The extensive application of ATM exchange has also brought new challenges to network monitoring and management of the switching network. Third, the development of the virtual local area network (VLAN) exchange technology has also accelerated the application speed of new exchange technology (VLAN). By dividing enterprise networks into virtual network VLAN network segments, network management and network security can be strengthened, and unnecessary data broadcasts can be strengthened. In the shared network, a physical network segment is a broadcast domain. In the switching network, the broadcast field can be a virtual network segment with a set of any selected second-layer network address (MAC address). In this way, the division of the working group in the network can break through the geographic location limits in the shared network, and is divided according to management functions. This workflow-based packet pattern greatly enhances the management functions of network planning and restructuring. Workstations in the same VLAN, whether they are actually connected to which switches, the communication between them is as if they are on a stand-alone hub. The broadcast only in the same VLAN can only be heard, and it will not be transferred to other VLANs, which can control unnecessary broadcast storm production. At the same time, if there is no route, different VLANs cannot communicate with each other, which increases the security between different departments in the corporate network. Network administrators can fully manage information exchanges between different management units within the enterprise between the VLANs. The switch is divided by the MAC address of the user's workstation. Therefore, users can freely move their office in the enterprise network, whether he access to the exchange network, he can communicate with other users in the VLAN.
