Detailed description of Ethernet switch performance comparison indicators

Rack-mounted switch, fixed configuration with/without expansion slot) rack-mounted switch is a slot type switch, which has good scalability and supports different network types, such as Ethernet, fast Ethernet, Gigabit Ethernet, ATM, licensing ring, and FDDI, but the price is relatively high. A fixed-configuration vswitch with expansion slot is a type of switch with a fixed number of ports and a small number of expansion slots. This type of switch can support other types of networks on the basis of a fixed-port network, the price is centered. Fixed configuration vswitches with no expansion slot support only one type of network, but the price is the cheapest.

Configuration:

Rack slots-the maximum number of modules that a rack-mounted switch can insert.
Number of expansion slots-refers to the maximum number of modules that can be inserted by fixed configuration switches with expansion slots.
The maximum number of switches that can be stacked in a stack unit. This parameter specifies the maximum port density available in a stack unit.
The minimum/maximum number of 10 M Ethernet ports is the minimum/maximum number of 10 M Ethernet ports supported by a vswitch.
The minimum/maximum number of M Ethernet ports is the minimum/maximum number of M Ethernet ports supported by a switch.
The minimum/maximum number of M Ethernet ports refers to the minimum/maximum number of M Ethernet ports that a switch can connect.

Supported network types:

Generally, fixed configuration switches with no expansion slot support only one type of network. Rack-mounted switches and fixed configuration switches with expansion slots support more than one type of network, for example, Ethernet, fast Ethernet, Gigabit Ethernet, ATM, licensing ring, and FDDI are supported. The more network types A vswitch supports, the higher the availability and scalability.

Maximum number of ATM ports-ATM is the asynchronous transmission mode. The maximum number of ATM ports is the maximum number of ATM ports supported by an ATM switch or a multi-service multi-function switch.
Maximum number of SONET ports-SONET is the abbreviation of Synchronous Optical Network. It is a high-speed synchronization Network specification with a maximum speed of 2.5 Gbps. The maximum number of SONET ports of a vswitch is the maximum number of downstream SONET interfaces of the vswitch.
Maximum number of FDDI ports: the maximum number of FDDI ports supported by an FDDI switch or a multi-service multi-function switch.
Backplane throughput bps)-also known as backplane bandwidth, is the maximum amount of data that can be transferred between the vswitch interface processor, interface card, and data bus. The higher the bandwidth of the backboard of A vswitch, the stronger the ability to process data, but the design cost will also rise.
Buffer size-sometimes called the packet buffer size, is a queue structure that is used by the switch to coordinate the speed matching problem between different network devices. Burst data can be stored in the buffer until it is processed by a slow device. The buffer size should be moderate. A large buffer space will affect the forwarding speed of data packets in normal communication, because a large buffer space requires a little more addressing time) and increase the device cost. A small buffer space is prone to packet loss errors when congestion occurs. Therefore, adding an advanced buffer scheduling algorithm to an appropriate buffer space is a reasonable way to solve the buffer problem. Pay attention to the following points for the network trunk device:

Whether each port has an independent buffer space, and the working status of the buffer space does not affect the buffering status of other ports;
Whether the module or port is designed with independent input buffer, independent output buffer, or input/output buffer;
Whether there are a series of buffer management scheduling algorithms, such as RED, WRED, RR/FQ and WERR/WEFQ.

Max MAC address table size-each port or device connected to the LAN needs a MAC address, which is used by other devices to locate a specific port and update the route table and data structure. The MAC address is 6 bytes long and is allocated by IEEE. It is also called a physical address. The MAC address table size of a device indicates the maximum number of nodes that can be connected to the device.

Maximum number of power supplies-generally, the core equipment provides redundant power supplies. After a power supply fails, other power supplies can continue to provide power, without affecting the normal operation of the equipment. When multiple power supplies are connected, pay attention to the use of multiple mains supply. In this way, when one line fails, other lines can still provide power.

Support protocols and standards-generally refer to the networking specifications and equipment standards formulated by the International Organization for Standardization. You can classify the network models at Layer 1st, layer 2nd, and layer 3rd as follows:

Layer 1: EIA/TIA-232, EIA/TIA-449, X.21, EIA530/EIA530A interface definition.
Layer 3: 802.1d/SPT, 802.1Q, 802.1p, and 802.3x.
Layer 3: IP, IPX, RIP1/2, OSPF, BGP4, VRRP, and multicast protocol.

The routing information protocol (RIP-RIP) is a distance vector protocol that uses the number of hops as the Metering Standard. RIP is widely used in the routing selection of global interconnection networks. It is an internal gateway protocol provided by the UNIX Berkeley standard distribution system. Ip rip is defined in RFC 1058 and RFC 1723.

RIP2-is the latest enhanced version of RIP. It allows RIP packets to contain more information and provides a simple authentication mechanism. It is described in RFC 1723.

Open Shortest Path First version 2nd (OSPFv2)-it is the second version of OSPF. OSPF is a connection status routing protocol, which is an IETF of an interconnected network engineering task group. The internal gateway protocol workgroup is designed for IP addresses, it is the next link State hierarchical routing algorithm of RIP in the Internet communication body. OSPF features include least-cost routing, multi-path, and load balancing. OSPF was developed from an earlier version of the IS-IS protocol and has two main features: first, the Protocol IS open, for example, RFC 1247 has OSPF rules. Second, OSPF is built on the SPF algorithm. SPF is also called DIJKSTRA algorithm. It is named after the founder of this algorithm.

The Border Gateway Protocol BGP--BGP is used to replace the EGPExterior Gateway Protocol) Inter-Domain Routing Protocol. The accessibility of information exchange between BGP and other BGP systems is defined by RFC 1163. BGP4 is the fourth version of BGP. It supports CIDR and uses the Routing aggregation mechanism to reduce the size of the route table.

The classless Inter-Domain Routing CIDR--CIDR is a routing clustering technology supported by BGP4. CIDR allows multiple routers to form a route group to reduce the routing information of the core router load. Based on CIDR, several IP networks can be separated from groups as independent and large entities.

Internet Group Management Protocol IGMP--IGMP is short for Internet Group Management Protocol. An IP host is used to report its members to an adjacent multi-view router. A multi-view router sends an IGMP query packet to the connected local network. A host Member of a multi-object group sends an IGMP Report of the multi-object group to respond to a request. The MnS router is responsible for forwarding the MNS from one MnS group to all other networks with members in this group.

Distance Vector Multicast Routing Protocol DVMRP--DVMRP is the gateway protocol of the interconnected network, basically based on RIP, can realize the typical Detection Mode of IP multi-object mechanism. DVMRP uses IGMP to exchange route data with neighboring points.

Open Shortest Path-first Multicast Routing Protocol MOSPF-multi-object OSPF is used for the Inter-Domain multi-object routing protocol of the OSPF network. The extended form can be used for basic OSPF single-host protocol to support IP multi-destination routing.

Protocol-Independent Multicast Protocol PIM--PIM is the abbreviation of Protocol Independent Multicast. It is a multi-object transmission Routing Structure, which can add IP multi-object transmission routing to the existing IP network. PIM is an independent route protocol for single-host transmission. It can be operated in two modes: intensive mode and loose mode. In PIM intensive mode, packet groups must forward data to all output interfaces until cropping and removal occurs. In dense mode, the receiver is dense, and it is assumed that the downlink network is ready to accept the data packets forwarded to it, and these data packets may be used. The cost of intensive mode is its inherent diffusion behavior. In the PIM loose mode, try to restrict data transmission, so that the number of routers that receive data in the network is minimized. In the loose mode, receivers are widely distributed, and it is assumed that the downlink network does not have to use sent data packets. The cost of using the loose mode is to explicitly combine the periodic refresh of packets and the requirements for RP convergence points.

Resource Reservation Protocol RSVP--RSVP is short for Resource Reservation Protocol. This Protocol supports Resource Reservation across IP networks. Applications running on the IP terminal system can use the RSVP protocol to indicate the attributes of the data packet stream to be received by other nodes, such as bandwidth and maximum burst. RSVP depends on IPv6.

802.1p priority mark and multiple queues-the IEEE 802.1p standard is used to classify various network applications and information flows. It ensures that key commercial applications and information flows with high time requirements are transmitted first, while taking care of low-priority applications and information flows to get the required services. This standard is essential for financial services, document processing, network management, integrated sound and data applications, video conferencing and Distributed Video Image teaching applications.

Routing: A process of finding the path of the target host is called routing. In a large network, a data packet passes through many intermediate paths before it reaches the target master, so the routing is complex. The implementation of a route in a vro relies on static protocols or dynamic routing protocols.

Supports layer-4 Switching-the so-called layer-4 switching is based on the exchange technology integrated with the routing technology, so that the switch can forward packets at line rate. A 3rd-layer switch is equivalent to a high-speed LAN router. Effective control using layer-3 switches
Broadcast storm, Spanning tree loop, and IP address restriction.

Supports layer-4 to layer-7 switching-a multi-layer switch is a type of switch that can filter and forward data packets based on MAC addresses and network addresses. It is an intelligent subset of LAN switches. Multi-layer devices can understand the application of the transmitted data packets. Therefore, multi-layer switching provides application-level control, that is, supports Security filtering and provides specific QoS policies for application streams.

Multi-protocol routing-multi-protocol routing is supported in a hybrid network environment of Multi-Protocol IP, IPv6, AppleTalk, and IPX, different heterogeneous networks can communicate with each other.

Supports protocols other than IP addresses-in addition to the IP protocol, protocols such as AppleTalk, DECnet, IPX, and NETBEUI are also supported.

Support routing cache-see "buffer size.

The maximum number of Route tables is supported. Route tables are stored in routers or other Internet devices to track specific network targets. The maximum number of Route tables is the maximum number of routes stored in a route table.

VLAN: a virtual LAN configures a group of devices on the LAN to communicate on the same line, but in fact they are in different network segments. A VLAN is an independent broadcast domain that effectively prevents broadcast storms. The VLAN is configured flexibly because it is based on logical connections rather than physical connections. Now, whether a vswitch supports VLAN is an important parameter for measuring the performance of a vswitch. When dividing VLANs, there are port-based, MAC address-based, layer-3rd Protocol-based, and subnet-based. 802.1Q is a VLAN standard. devices of different vendors can interconnect and divide VLANs as long as they support the 802.1Q standard. Based on layer-3 switching, VLANs can also communicate with each other.

Maximum number of VLANs-this parameter reflects the maximum number of VLANs supported by a device. The maximum number of VLANs supported by a switch is more than 1024.

Network Management: a network administrator manages network resources in a centralized manner through network management programs, including configuration management, performance and accounting management, problem management, operation management, and change management. The management level supported by a device reflects the manageability and operability of the device.

Network management supported

Common network management types include:

IBM Network Management NetView)
HP OPENVIEW
SUN Solstice Domain Manager
RMON Management
SNMP Management
WEB-based management
Management Interface
Network management operations. CLI and GUI are supported. This parameter reflects the operability and availability of the device.
Supported port Images
The so-called port image uses SPAN to accept port monitoring and redirects the source port (mirror data port) to the target port. Many vswitches are implemented using the Set SPAN or SPAN command.
Supported RMON and RMON2 groups
Supports RMON groups:
Statistics)
History)
Warning (Alarms)
Host (Hosts)
Host Top N (Host Top N)
Session Matrix)
Filter)
Packet Capture)
Events)
RMON2 provides monitoring capabilities at or above the link layer. RMON2 groups:
Protocol Directory)
Protocol Distribution)
Address Mapping)
Network Layer Host)
Network Layer Matrix)
Supported packet filtering methods:
Define the access list ACL)
Source Address Filtering
Destination Address Filtering
Protocol type field packet filtering
B programmable field filtering
QoS Service Quality of Service (QoS) is the performance measurement of the transmission system, reflecting the transmission Quality and Service availability. It is mainly ensured by RSVP and 802.1P.

Supports policy-based 2nd-layer switching-The 2nd-layer function not only supports data frame processing, but also supports industry standards, such as IEEE 802.1d Spanning Tree Algorithm) and IEEE 802.1p priority queue control) and IEEE 802.3x Traffic Control for congestion.

Maximum number of priority queues per port-port queue is a technology developed to solve network congestion problems. The maximum number of priority queues per port reflects the capability of devices to handle network congestion.

Supports policy-based 3rd layer switching-The 3rd layer function is not only limited to packet processing, but also supports three types of functional protocols: data protocols, such as IP, IPX, DECnet, and AppleTalk; routing protocols, such as RIP, r00002.0, OSPF, and BGP4; multicast protocols, such as IGMP, DVMRP, and MOSPF.

Supports policy-based application-level QoS-implements access control ACLs for data streams for different applications based on header information of layer 4th), service quality assurance QoS), and bandwidth management, and various control and service policies.

The ability to implement policies on data streams mainly includes:

Traffic classification, traffic control, and the ability to provide different service levels for data streams generated by different applications.
Minimum/maximum bandwidth allocation supported
Bandwidth allocation is a way for devices to support policy management. Dynamic Bandwidth Allocation can significantly save bandwidth without affecting data transmission quality.

Redundancy-redundancy emphasizes the reliability of the device, that is, the device is not allowed to have a single point of failure.

Redundant Component Management Card, switching structure, interface module, power supply, cooling system)

The device should have component-level backup, such as redundancy for power supplies and chassis fans. When a part fails, other parts can continue to work without affecting the operation of the device. However, such components may not be hot-swappable.
Heat Exchange Component Management Card, switching structure, interface module, power supply, cooling system)
For management engines and exchange array modules that provide key services, they require not only redundancy, but also "automatic switching" to ensure the integrity of redundancy, when such a part fails, the redundant part can take over the work to continue the operation of the device. This ensures the reliability of the device.

Support Port Link aggregation protocol-link aggregation refers to connecting several ports of a device to the same port of another device and requiring the same media) to provide several times of bandwidth. Link aggregation is managed by the link aggregation protocol. When a link fails, the link aggregation protocol coordinates other links to continue working. This parameter reflects the redundancy and scalability between devices.

Server Load balancer-this parameter refers to the ability of a vro or layer-3 Switch to allocate service traffic to those network ports with the same distance as the destination address during the routing process. Good load balancing algorithms use line speed and reliability information. Server Load balancer increases the utilization of network segments and thus improves the effective network bandwidth.