What Is Latency
In a communications network, latency refers to the time it takes for original data to go through a series of processing operations such as encoding on a forwarding device, transmission from the transmit end over transmission links, and reception and decoding on the receive end (destination). The network latency consists of the following parts:
Sending latency: The period from the time when the original data enters the forwarding device to the time when the original data completely enters the transmission medium, depending on the data volume/channel bandwidth and processing performance of the devices.
Transmission latency: The period from the time when the data is sent from the transmit end to the time when the data is completely received by the receive end, depending on the transmission distance and transmission medium.
Processing latency: The period taken by the receive-end device to decode and restore the received information to the original data, depending on the processing performance of the device.
WDM/OTN Latency Distribution
The WDM/OTN network is an optical transmission system that uses optical fibers as the signal transmission media. As shown in the preceding figure, transmission links and physical devices are the main factors affecting network latency in a typical WDM/OTN system.
Figure 1 Typical WDM/OTN network
The latency of physical links is mainly introduced by line fibers and dispersion compensation fibers (DCFs).
- The latency introduced by line fibers refers to the time it takes for optical signals to transmit over the line fibers.
- The latency introduced by DCFs refers to the time it takes for optical signals to transmit over the DCFs.
The latency of devices is mainly introduced by electrical- and optical-layer units.
- The latency introduced by optical-layer units refers to the latency generated when they process optical signals.
- The latency introduced by electrical-layer units refers to the latency generated when electrical signals converted from optical signals at the receive end undergo a series of operations on the board, such as mapping and forward error correction (FEC) encoding and decoding.
WDM/OTN Latency Performance Optimization Technologies
According to the preceding analysis, the WDM/OTN network latency is mainly introduced by devices and line fibers. The fiber transmission latency accounts for more than 90% of the network latency, followed by device components. Huawei WDM/OTN technology optimizes networking and optical-layer and electrical-layer devices to reduce the end-to-end (E2E) transmission network latency and ensure the minimum network latency.
Simplified network architecture design, one-hop service transmission, and reduced latency
The simplified network architecture design reduces forwarding nodes, constructs a one-hop transmission network, and reduces the system latency. For example, the conventional ring or chain topology is optimized to a full-mesh topology during backbone network planning and design. On the metro network, WDM/OTN devices are shifted downwards to CO nodes.
Optical-layer optimization:
Leverages the coherent communication technology, DCM-free, to eliminate the extra latency caused by DCFs.
Replaces the DCF-based dispersion compensation modules with the FBG-based dispersion compensation modules.
Leverages the advanced optical-layer technologies of ROADM/OXC to implement optical-layer pass-through and switching, reducing the number of OEO times.
Replaces EDFAs with intelligent Raman amplifiers. These Raman amplifiers do not need the erbium-doped fiber as the medium. This avoids extra latency caused by the EDFA, effectively extends the all-optical transmission distance, and reduces the number of electrical regeneration sites.
Electrical-layer optimization:
Optimizes FEC algorithm performance, increases the transmission distance, and reduces the number of electrical regeneration sites.
Flexibly sets the number of FEC levels and reduces the latency penalty.
Sets different encapsulation modes for various features.
WDM/OTN Latency Management Solution
Latency management is a comprehensive application based on latency measurement and data estimation of network links and nodes to provide network or service latency that can be sensed, sold, committed, and guaranteed.
Measurable Latency
E2E latency indicators are essential to verification of service level agreements (SLAs) between users and network carriers, especially to latency-sensitive services. Based on the real-time network latency monitoring data, network carriers establish dynamic network latency models and adjust the network transmission policies by predicting latency changes. Therefore, how to obtain the latency data of the transmission link is particularly important.
Traditional methods for measuring the link latency have shortcomings, such as inaccurate precision and service interruption during measurement. For example:
Externally connected meters: Services are interrupted during the measurement.
Estimation: The transmission latency is estimated based on the lengths of the fiber links at the source and sink ends. This method ignores the latency introduced by the electrical layer and has low precision. In addition, the length of the fiber link is difficult to obtain.
To provide latency-sensitive customers with latency performance priority services, ITU-T G.709 adds the latency measurement function to support online real-time monitoring of network latency. In ITU-T G.709, this function is associated with ODUk PM-layer and TCM-layer overheads. The PM and TCM overheads are latency measurement overheads and located in row 2 and column 3 of the OTU frame. Bit 7 is related to the PM layer test, bits 1 to 6 are related to the TCM layer test, and bit 8 is a reserved bit (the default value is 0). The function is classified into ODUk PM-layer test (DMp) and ODUk TCMi-layer test (DMti). The latency at different TCMi layers can be tested based on the TCMi enable status.
The following figure shows the measurement process.
The initiator of latency measurement encapsulates services, inserts the latency measurement bytes into the ODUk overheads, starts the latency measurement, and records the current time.
The latency measurement bytes are transmitted along the links with the services. If an intermediate node exists, the intermediate node transparently transmits the latency measurement bytes and does not process them at the ODUk layer. When the latency measurement bytes reach the terminator of the latency measurement, the latency measurement bytes are transmitted to the initiator in the reverse direction.
The initiator receives the latency measurement bytes from the reverse transmission and compares them with the start time of latency measurement to calculate the round-trip latency from the initiator to terminator. Considering the consistency of WDM/OTN routes, half of the round-trip latency can be regarded as the one-way transmission latency.
This latency measurement is based on ODUk overhead bytes and does not affect services. Therefore, this function provides in-service latency monitoring. The measurement precision is μs-level, which can meet most service requirements. The latency measurement results include the electrical-layer processing latency of service boards and the optical-layer transmission and processing latency of transmission fibers, OA, DCM, and OADM. To ensure the test accuracy, you can perform multiple tests and take the average value.
