Is Cat5 running Gigabit Ethernet?

At present, most commonly used Ethernet Media are cables, supplemented by optical cables. Generally, 10 BASE-T Ethernet requires only three types of links. 100Base-T Ethernet requires five types of links to ensure its operation. In addition, a considerable number of five types of links can support BASE-TX applications, the cat5e chain can stably support 1000Base-Tx links. During the upgrade process from many users to 100Base-T and from a few users to 1000Base-Tx, the so-called "upgrade pain" problem occurs to varying degrees: The speed cannot meet the expected standard, even unable to enter the corresponding speed "status", except for the dangers of "fake and shoddy" products, the problems of construction technology and network work environment are also an important cause of "upgrade pain.

Take the network cable as an example. To stably support the speed of 100Base-T, the five links require that all link components should be applicable to five components. The link authentication test requires four parameters (TIA) to be tested) or six parameters (ISO11801). If 1000Base-Tx is supported, you need to add parameters such as PSACR and ELFEXT. One of these authentication test parameters is unqualified, which will affect the speed that the network actually supports to varying degrees. We will use several topics to introduce the network faults caused by media faults.

On a futures exchange, only one CIDR block can work normally after the network is transformed to a Gigabit Ethernet (backbone) network. The other 12 CIDR blocks do not work normally and data errors occur, the connection is often inexplicably interrupted. Each CIDR block is connected over 1 Gbit/s Ethernet, And the downlink CIDR block is the 100Base-T user port. At first, it was suspected that the platform or software running the system had a problem. After multiple reinstallation and setup, the problem still could not be solved, and the same system was applied normally in transaction networks elsewhere. Therefore, it is suspected that the cabling system is faulty, for example, the cable is unqualified or the interference signal is stringed in and the grounding system. Each CIDR block is connected by a pre-upgrade cable system without significant changes. Because the cabling system of computer networks adopts the standard five-class line scheme, according to the design standard of Gigabit Networks, four pairs of lines are used for full duplex operation and 5-level coding, the physical bandwidth occupied by the signal is exactly 100 MHz, so the five lines should be fully competent. Moreover, in general, the current traffic level of the futures trading network is far from fully loaded, low traffic. Re-use a professional cable Tester for strict certification tests, the display parameters are qualified and there is no pulse noise interference or grounding problems.

The so-called CIDR block is also because the quote server and the Transaction Server are both installed in this CIDR block. The workstation and server in This CIDR block can be connected to the Internet and work properly except for some sites, quote browsing and transaction cutover services are also smooth. Servers in other CIDR blocks are normally connected to each other in addition to individual workstations. The common feature is that they cannot be well connected to the CIDR Block of the quote server and the Transaction Server. When the system is upgraded, all the original cabling cables are retained, and all the cables pass tests. I wonder why?

There are many factors that cannot be connected, such as functional settings of network hardware devices, wiring system, operating platform installation and setup, installation and Setup of application software, and software conflicts. According to the user's situation analysis, all the sites in each CIDR block can work, Ping tests between CIDR blocks are unstable, and there is a certain loss rate, which makes the data exchange process more difficult, there are some interruptions. Based on past experience, it can be preliminarily determined that the fault may appear more likely in terms of network equipment setup and cabling system performance.

Connect the network tester F68X to the network segment (100 Mbps) of the server and the Transaction Server. observe that the network traffic to the server reaches an average of 12% in 5 minutes, and the frame verification error is about 11%, the collision rate is 1.7% (normal range ). Obviously, the percentage of errors in the verification of the FCS frame is too high. Check the error source and display that the percentage of errors in the production of the FCS frame on other network segment sites accounts for 97% of the total number of error frames. The error frame ratio of each CIDR block is not significantly different. Due to the widespread existence of a large number of FCS frames, it is very unlikely that all sites in each network segment have problems at the same time. F683 is used to send traffic to servers or sites in each network segment, observe that the port number of the switch rapidly increases with the increase of traffic, and the reflection of each site or server is basically the same. Start the ICMP Ping function of the network tester and calculate the test results of the sites, switches, routers, and other sites selected in each network segment. The performance is similar: ICMP Ping fault is about 96%, ICMP Monitor shows that the target cannot reach 91%. Test the same content in other CIDR blocks. the router and switch results of the CIDR blocks of the quote server and the Transaction Server are basically the same as those in the previous test. The difference is that the test results for vswitches, routers, and other network devices in other network segments are normal. The value is: ICMP Ping fault is 0%, all of which are accessible, the ICMP Monitor Target cannot reach 0%. The fault lies in the connection link between the server network segment and other network segments. Use FLUKE company DSP-4000 cable certification tester selection TIA Cat5n Channel UTP100 standard test, the display length is 25 meters, the link test is unqualified. Among them, the Return Loss RL and attenuation crosstalk are superior to those of ACR and other parameters. Use Cat5e, A Cat5e line with the same length, to start the system, except for individual sites in each network segment. During peak traffic monitoring, the average traffic in the highest network segment of the server is 3%. It can be seen that 12% of the traffic in the fault mainly comes from a large amount of resend frame traffic.

GE Ethernet can satisfy the "greedy" appetite of network users for large bandwidth applications, which is undoubtedly the next major development direction of the network. When selecting the cable type, designers of Gigabit Ethernet have considered a large number of existing Cat5 applications. Therefore, the physical bandwidth is 100 MHz. In this way, in principle, Class 5 lines can run Gigabit Ethernet. However, the actual statistical results show that there are still 1% ~ 5% of users cannot access the Internet or have intermittent connections and difficulties (some five links can even reach 30% ). That is to say, Gigabit Ethernet has stricter requirements on the parameters of Category 5 cables. As long as you perform strict Authentication Tests on five types of cabling systems, most sites can work online. A few sites may have difficulty accessing the internet due to small parameter margin. Parameters that have a greater impact include integrated near-end crosstalk ps next, integrated remote crosstalk ps fext, equivalent remote crosstalk ELFEXT, integrated equivalent remote crosstalk ps elfext, Return Loss RL, and attenuation crosstalk ratio ACR. Cat5n standard testing is required for Category 5 lines. This standard is designed for users who use category 5 lines to run Gigabit Ethernet. If all the tests comply with this standard, then, you can safely run Gigabit Ethernet with a Class 5 line system. In the new Cat5n standard, the return loss has a great impact on the system. In addition, due to the non-continuous impedance in cable matching, the shorter the cable link, the more prone to problems. In this example, because the cable length is 25 meters, although the attenuation crosstalk is lower than the ACR parameter, the return loss is more likely to cause this fault. (Note: Cat5n is the TSB95 certification test standard and is now integrated into the new TIA568B standard .)

Cat5n (TSB-95) standards can be selected for the certification test of Category 5 lines, so that it can be guaranteed to run Gigabit Ethernet network no problem. If Cat5e is used for cabling, there is no possibility of running Gigabit Ethernet. In order to minimize the cost and improve network performance, one of the simplest ways is to replace the defective individual links with Cat5e over five Cat5e lines. Note: It is best to replace the connection module together to ensure the installation quality of the link.

Following our advice, two weeks later the futures exchange will all the installation links with DSP-4000 cable tester selected TSB-95 standards for testing, all site number is 1300, 21 unqualified links. After switching to more than 5 types of cables for unqualified links, all the sites will work normally and the network status will remain excellent (the average traffic of this network is only 3% during peak hours ).