Cabling Area in the Data Center

With more and more functions of the data center, the scope of the computer room is also expanding. Many weak current systems are deployed in the data center, such as the communication interconnection between network element equipment, closed-circuit monitoring, environmental power monitoring, access control KVM and other subsystems. This research focuses on the communication interconnection subsystem between network element devices.

The main network element devices in the data center include hosts (servers), disk array storage, access switches, aggregation switches, core routing switches, etc., as well as patch panels (boxes) for optical (electrical) cables to connect these main devices. Wait. The communication interconnection between network devices constitutes the communication subsystem of the integrated wiring. The wiring of the communication interconnection subsystem is divided into main wiring area (MDA), horizontal wiring area (HDA), regional wiring area (ZDA) and equipment wiring area (EDA) according to functional areas. MDA is at the core layer of the data center network, including core routers, core switches, and core storage switches; HDA corresponds to the aggregation layer of the data center network, and the main device is the aggregation switch; ZDA is a regional network system, such as disk storage The storage area network (SAN) formed between the array and the access switch is an area network. The EDA is at the access layer, and the main equipment is the host, server, storage, KVM, etc. The equipment wiring is connected in the cabinet.

1) In the equipment distribution area (EDA), the server is connected to the downstream port of the access layer switch, and the upstream port is connected to the aggregation layer switch. The transmission rate of the upstream port is an order of magnitude higher than that of the downstream port, such as the downstream port is 1 Gbit/ s, the upstream port is 10 Gbit/s. The uplink port of the access layer switch is an optical port (optical module interface), and the downlink port can be connected to the server using an optical port. If the distance is less than 30 m, an electrical port can also be used, and Cat.6A twisted pair is used for connection.

2) In the horizontal distribution area (HAD), the downlink port of the aggregation layer switch is connected to the uplink port of the access layer switch, and the uplink port is connected to the core layer routing switch, and its bandwidth capacity must be more than three times that of the downlink port to ensure no occurrence of network congestion. That is to say, if the downstream port of the aggregation layer switch is 10 Gbit/s, the upstream port must reach more than 30 Gbit/s, generally 40 Gbit/s optical port configuration.

3) In the main distribution area (MDA), the main function of the core layer is to realize high-speed exchange of data packets, it is the final convergence point and processing point of all traffic, and provides access to the WAN link. Core layer routing switches generally use high-performance multi-layer modular switches. Currently, core layer routing switches increase bandwidth by adding redundant links, and routers can provide load balancing functions for multiple links and paths, thereby improving data forwarding efficiency. Currently, the total bandwidth reaches 100 Gbit/s.

Optical Modules in Data Centers

The optical module is an important component of the optical fiber communication system, which is used to realize the generation, transmission and reception of optical signals. The digital bit rate of optical modules used in data centers has changed from 1G before 2010 to the commonly used 10G. Currently, 40G and 100G are popular, and 200G and 400G will be used in the near future. Since most of the devices in medium and large data centers are within 500 m, multi-mode fibers (OM3, OM4) are generally used for interconnection between devices, and VCSEL lasers are used as the light source in the optical module as the light source and PIN is used as the optical receiver. . At present, the optical interconnection between the server and the access switch in the access layer mainly uses 10G optical modules, which can be realized by single-wavelength dual-fiber duplex optical modules. 10GBASE-SR, 10GBASE-LR, has the characteristics of small size and hot swap.

1) At present, 40G ports are mainly used from the upper port of the access switch to the port of the aggregation switch. In 2013, the 40G optical module of QSFP+ was deployed on a large scale to form a 40GBASE-SR4 Ethernet. There are 4 optical transmitters in the optical module. And 4 optical receivers, MPO/MTP connectors of 12-core multimode fiber are inserted on the optical interface of the optical module, of which 4 cores are sent and 4 cores are received, and 4 cores are vacant.

 2) From the port on the switch on the aggregation layer to the core layer routing switch, it is more popular to use 100G ports. At the beginning, CPF2 was used to encapsulate a 100G optical module. This module meets the requirements of the 100G Base-SR10 specification and provides 10 parallel data channels. , each channel transmits 10G, using 20-core OM3 or OM4 multimode fiber for parallel transmission, the module interface adopts 24-core MPO plug-in interface, 10-core sending and 10-core receiving, and the 4-core on the side is not enabled.

3) QSFP28 is the latest 100G Ethernet component because it is smaller in size, lower in power consumption and uses less fiber than CPF2. Therefore, the network form of 100G BASE-SR10 has gradually withdrawn from the market. Since 2017, QSFP28 optical modules with 100GBASE-SR4 interface are often used in data centers. It adopts 8-core OM4 multi-mode fiber for parallel transmission, and the module interface adopts 12-core MPO, 4-core sending and 4-core receiving, each channel supports 25G, and the middle 4 cores are not enabled, which is consistent with the 40GBASE-SR4 interface.