Traditional data centers rely on 10G network architecture. However, in order to meet the large-scale deployment requirements of services such as AI, deep learning Xi, and big data computing, the next-generation data center architecture is transforming to a 25G and 100G network architecture. In China, we have seen Internet leaders such as BAT achieve large-scale deployment of this high-bandwidth network architecture.
In the process of building 25G and 100G data centers, the demand for 100G optical modules has increased significantly, and their proportion in network construction costs cannot be ignored. So, what are the standards for 100G optical modules?How should the different standards be applied?This article will provide you with a concise overview of the standards, packaging forms, and application scenarios of 100G optical modules in data centers, aiming to provide an accurate reference for your selection.
A 100G optical module is an optical module with a 100Gbps optical signal transmission rate.
Before the optical transceiver, we must first understand its standardization organization. Two key organizations are responsible for the definition and specification of optical modules: IEEE and MSA (Multi Source Agreement). The two organizations complement each other and draw on each other's standards. MSA has formulated a number of standards for 100G optical modules, including 100G PSM4 MSA, 100G CWDM4 MSA, and 100G Lambda MSA.
In order to meet the requirements of 100G interconnection over different distances, industry organizations such as IEEE and Multi-Source Protocol (MSA) have formulated multiple standards for 100G optical modules. Among these standards, the PSM4 and CWDM4 standards developed by the MSA industry organization are particularly applicable to the mainstream 100G QSFP28 optical modules in the current market.
The following ** lists some common 100G optical module standards:
The package forms of 100G optical modules mainly include CFP, CFP2, CFP4, CXP, and QSFP28.
CFP is the first packaging format to be introduced, the 100GBASE-SR10 standard for short-distance transmission and the 100GBASE-LR4 standard for long-distance transmission. Since the electrical interface capability of the first generation CFP was limited to CAUI-10, a built-in gearbox was required to convert 10x 10Gbps to 4x 25Gbps electrical signals. With the upgrade of the electrical signal specification to CAUI-4, the second-generation CFP (including CFP2 and CFP4) no longer needs a built-in gearbox in long-distance transmission schemes. However, due to the large size of CFPs, with the improvement of the integration of optical modules, the subsequent development trend is to reduce the size and power consumption, thus evolving to CFP2 and CFP4.
As the demand for high density in data centers continues to increase, the more miniaturized packaging formats CXP and QSFP28 have emerged. Compared with CFP2, CXP optical modules have a smaller size and can meet the needs of high-density cabling, making them a viable solution for short-distance transmission. The QSFP28 optical module is smaller in size and consumes less power than CXP. Its small size allows the switch to achieve higher port density, typically with 36 100G interfaces per board. At present, QSFP28 has become the mainstream packaging format for 100G optical modules in data centers.
100G short-range solution
The 100G short-haul transmission solution mainly relies on two models: 100GBASE-SR10 and 100GBASE-SR4. Compared with 100GBASE-SR10, 100GBASE-SR4 optical modules show significant advantages in terms of device number, cost, module size, and power consumption. This is reflected in a reduction in the number of devices, a reduction in cost, a reduction in module size, and a decrease in power consumption. The reduced module size allows the switch to deliver a higher 100G interface density per 1U of space.
In view of the above advantages, 100GBASE-SR4 has gradually replaced 100GBASE-SR10 and has become the mainstream 100G short-range optical module standard.
The 100G QSFP28 SR4 optical transceiver is mainly used for short-distance connectivity in data centers and enterprise network environments. In most cases, the 100G QSFP28 SR4 transceiver is considered ideal for short-range 100G direct connectivity applications and upgrades from 25G to 100G. Its superior performance and economy make it occupy an important position in the field of high-speed data transmission.
100G QSFP28 SR4 optical module 100G direct connection solution.
100G QSFP28 SR4 optical module 25G-100G upgrade solution.
100G mid-range solution
In medium-distance transmission scenarios, the transmission distance range is typically defined from 100 meters to 2 kilometers. 100GBASE-SR4 and 100GBASE-LR4 are two common models of 100G optical modules. However, in the interconnected environment inside large data centers, 100GBASE-SR4 cannot meet all interconnection requirements due to its limited transmission distance, while 100GBASE-LR4 is not economical due to its high cost.
To this end, the Multi-Source Protocol (MSA) has responded to the needs of the market with the introduction of mid-range interconnection solutions - PSM4 and CWDM4 optical modules, which are technological innovations to meet this challenge.
The QSFP28 PSM4 optical transceiver is ideal for low cost transmission distances of less than 500 meters. The most common application method is to achieve 100G-100G direct connection. During the implementation, you only need to plug two 100G QSFP28 SR4 optical modules into the corresponding ports of the two 100G switches, and then use OM3 or OM4 multimode fiber patch cords to connect the two modules to complete the configuration.
100G QSFP28 PSM4 optical module 100G direct connection solution.
On the other hand, 100G QSFP28 CWDM4 optical transceivers are often used for 100G interconnects of up to 2 km in data centers and enterprise networks. The cabling process is simple, you only need to plug two 100G QSFP28 CWDM4 optical modules into the corresponding ports of the two 100G switches, and then use LC duplex fiber patch cords with fiber adapter panels or fiber distribution boxes to achieve 100G interconnection. Among them, the fiber optic adapter panel or fiber optic distribution box can be installed on the fiber optic distribution frame, and this wiring method is especially suitable for structured cabling systems. Through this design, CWDM4 optical modules can provide a more flexible and economical mid-distance interconnection solution while ensuring transmission performance.
100G QSFP28 CWDM4 optical module 100G interconnection solution.
100g long-range solution
In long-haul transmission scenarios, 100G QSFP28 LR4 optical modules are often regarded as an ideal solution for 100G direct connection and interconnection.
For the 100G-100G direct connection scheme, the connection process is relatively simple. Simply plug two 100G QSFP28 LR4 optical modules into the corresponding 100G switch ports, and then use an LC fiber patch cord to connect the two modules to complete the long-distance 100G direct connection configuration.
Direct connection solution for 100G QSFP28 LR4 optical modules.
In 100G-100G interconnect scenarios that require high-density cabling, the implementation of the solution is relatively complex. As shown in the figure below, the 100G QSFP28 LR4 optical module can be efficiently interconnected to meet the requirements of high-density cabling by using optical fiber distribution boxes, MTP fiber adapter panels, and MTP backbone jumpers. Among them, the role of the adapter panel and fiber optic distribution box is to simplify cable management and improve the cleanliness and maintainability of wiring.
Direct connection solution for 100G QSFP28 LR4 optical modules.
From the above description of the 100G optical module package, we can understand that the 100G QSFP28 optical module and the 40G QSFP+ optical module are the same in package size, and both contain 4 integrated transmit and receive channels. We have already talked about how to use the QSFP28 port to achieve 10G 25G 40G transmission, so is it possible to plug the 100G QSFP28 optical module into the 40G QSFP+ port?
Theoretically, plugging a 100G optical module into a 40G port may cause an unstable connection or failure to function properly. This is because the transmission rate of a 100G optical module exceeds the design rate of a 40G port, which may lead to a rate mismatch.
To ensure the normal progress of data transmission and communication, the rates of optical modules and ports need to be matched. If you insert an optical module with a rate of 100 Gbit/s into a port that only supports 40 Gbit/s, the following problems may occur:
1.The connection is unstable: Frequent drops or interruptions may occur.
2.Reduced performance: Even if you stay connected, performance can be significantly degraded due to rate mismatch.
3.Incompatibility: Some devices may reject or fail to recognize a combination of transceivers and ports that do not match the rate.
To ensure the normal operation of the device, it is recommended to use an optical module that matches the port rate. If you need to connect a 40 Gbit/s port, select an optical module that supports the 40 Gbit/s rate. In practice, it's best to consult device manuals, technical specifications, or contact the device manufacturer for accurate compatibility information.
To sum up, it is recommended that you refer to the following standards for how to choose 100G optical modules for internal interconnection in 25G 100G data centers
100G short-range interconnection (Tor-Leaf) scenarios that do not exceed 100 meters use a 100GBASE-SR4 QSFP28 optical module
100G medium-range interconnection (leaf-spine) scenarios from 100 m to 500 m, using 100G PSM4 QSFP28 optical modules
100G medium- and long-range interconnection scenarios (Leaf-Spine and Spine-Core) from 500 m to 2 km use a 100G CWDM4 QSFP28 optical module
For long-distance interconnection scenarios (core-MAN) over 2 km, a 100GBASE-LR4 QSFP28 optical module is used.
In practical applications, the selection of appropriate 100G optical modules should consider factors such as transmission distance, cost, power consumption, port density, and device compatibility to ensure the efficient and stable operation of data center networks.