Support for New-Generation MSTP Performance and Storage
Information technology has entered an era of rapid development, with a wide variety of personalized services emerging one after another, from mobile communications to broadband services, from multimedia messaging to VOD on demand, all of which continue to increase the pressure on the underlying network. As the most direct support of the underlying network, the transport network should not only continuously expand the access layer of the original network, but also promptly introduce large-capacity scheduling equipment at the core layer to meet users' explosive demand for bandwidth.
The new-generation 10G MSTP equipment has not only achieved significant improvements in cross-connect capacity and service support, but also further matured in equipment performance and networking capabilities, and is bound to become the core force of the telecom transport network in the near future.
The system slot configuration can basically represent the manufacturer's design philosophy for its 10G product system architecture. The initial 10G MSTP systems were generally designed with a vertical plug-in board and upper/lower layered architecture, mainly to divide the interface area and processing area, and to facilitate the insertion and removal of various boards. Due to the limitation of backplane bandwidth, the 10G trunk boards usually had fixed slots, while the cross-connect boards were located at the center of the entire subrack. With the advancement of cross-connect technology, later 10G MSTP systems de-emphasized the distinction between trunk and tributary concepts.
Recently, there has been a trend toward miniaturization of 10G MSTP, and horizontal plug-in board subracks from individual manufacturers have gradually begun to appear. To maximize space savings, such systems often have 1 to 2 built-in 10G optical interfaces on the cross-connect board, but public boards such as the cross-connect board still have fixed positions, and the equipment needs to focus on solving heat dissipation issues.
Some domestic manufacturers have proposed a new type of cross-connect matrix, called fully interconnected cross-connect. By integrating cross-connect chips onto the optical interface boards and using space division to achieve full bus interconnection of service boards, they effectively form a 2-stage CLOS matrix.
New 10G MSTP systems generally have a low-order cross-connect matrix built into their high-order cross-connect matrix, but may not necessarily have sufficient service board slots, so their ability to expand low-order access and high-order cross-connect capacity requires attention. In terms of optical transceiver modules, the industry has currently achieved hot-swappable GE interfaces on 10G MSTP through the GBIC method, but attitudes toward hot-swapping other interfaces are inconsistent.
In addition, 10G MSTP has many commendable aspects in terms of universality and maintainability. Generally, foreign manufacturers can use various low-rate optical interface boards on both 2.5G and 10G platforms, and some manufacturers can even use all optical interface boards except 10G across their entire product line. Most domestic manufacturers basically do not support board universality, and only a few support mixed insertion of data boards.
Driving Network Performance Improvement
Network Capacity
In principle, the greater the cross-connect capacity of SDH equipment, the greater its access capacity.
The cross-connect board configuration of some manufacturers can be upgraded as the required access capacity increases. Although this saves operator investment in the initial stage, in the long run, upgrades often require reconfiguring all cross-connections, increasing the workload of operations and maintenance personnel and posing network security risks.
Redundant Backup
There are two main types of redundant backup for 10G MSTP equipment: 1+1 backup and 1:1 backup. 1+1 backup means that both the protection board and the working board are in operation, and no matter which board fails, the normal operation of the equipment is not affected; 1:1 backup means that while the working board is running, the standby board continuously monitors the primary board. Once a fault occurs on the primary board slot, the standby board slot immediately begins operation, and the network management system instantly switches the signal to the standby board slot while issuing an alarm signal.
Network Management Capability
The current transport network management system adopts a layered architecture according to the TMN structure, forming a three-layer system architecture: element-level network management, subnetwork-level network management, and network-level network management systems.
Network management at each level mainly provides Q interfaces and F/f interfaces, and most manufacturers also support CORBA interfaces, allowing unified network management of MSTP products at different levels from the same equipment manufacturer on one platform, and even unified management of their DWDM platforms. However, unified management of products from different manufacturers is not possible and requires third-party intervention. The layering of network management in most manufacturers is not clear-cut: element-level network management often has some subnetwork-level network management functions, and subnetwork-level network management often has some network-level management functions.
Most manufacturers define different numbers of network elements manageable by a single network management system, and some manufacturers adopt the concept of equivalent network elements, such as a 10G equipment being equivalent to 4 equivalent network elements; other manufacturers have proposed logical network elements for definition. However, these methods cannot avoid a drawback: they cannot distinguish different network management requirements under the same type of equipment.
"3+2" in Services
ATM Processing
ATM service requirements mainly come from current ADSL and future 3G.
Regarding the ATM requirements of UMTS services, although some NodeBs provide IMA interfaces, considering that 10G MSTP is located at the network core layer, direct processing of low-level services should be reduced, so it is not recommended to implement built-in reverse IMA functionality.
Some manufacturers access ATM through ordinary SDH optical ports, while others provide ATM interfaces through ATM processing boards. It should be noted that for the division of minimum aggregation granularity, domestic manufacturers are limited to VC4, while some foreign manufacturers can achieve VC12.
As a new derivative technology, some manufacturers support VP-Ring, but the practicality of this technology is worth further discussion. A few manufacturers also support S-PVC connections and PNNI signaling, but these technologies do not show particular advantages in ring network structures.
L2 Adaptation and Switching
Domestic manufacturers often distinguish between transparent transmission and switching for data boards, while foreign manufacturers generally do not. Most manufacturers support aggregation from FE to GE and FE to FE. For LAN-LAN aggregation, implementation on a single board is not a major issue, but for WAN-LAN aggregation, due to differences in aggregation ratios and virtual concatenation methods, implementation varies among manufacturers. Some manufacturers also provide an aggregation board that uses a simple "multiple WAN nodes sharing several VC-12" method for aggregation, which differs from L2 switching and should be distinguished during use.
Embedded MPLS
To solve the end-to-end service QoS problem across RPR rings, MPLS technology has been introduced into 10G MSTP. By establishing end-to-end LSPs, MPLS ensures sufficient guarantee for user services from start to finish.
Some manufacturers' approach is to add MPLS frames after L2 switching and before RPR. This method mainly uses the control functions of MPLS for user isolation, with overall performance still relying primarily on RPR. Other manufacturers' approach is to abandon RPR, directly encapsulate into MPLS frames after L2 switching, and then rely on the EOMPLS method, using MPLS fast forwarding and label recognition capabilities to strengthen support for Ethernet and VLAN. MPLS protection can adopt a false trigger mechanism, quickly performing LSP switching by modifying LSA information.
Some manufacturers exaggerate MPLS's solution to the VLAN address overlap problem, but this is not the greatest benefit brought by MPLS, as the VLAN Stack mechanism can also solve this problem. We believe that the greatest advantage of 10G MSTP with embedded MPLS is its support for dynamic collaboration of LDP signaling, enabling MSTP to interoperate with MPLS label routers in the near future, achieving seamless user switching within MPLS networks and extending the scalability and security of MPLS networks.
Currently, a few manufacturers already support Martini-model MPLS, and some manufacturers have implemented MPLS-VPN based on the MP-BGP model on 10G MSTP. From a technical perspective, the latter has greater practical significance in the network.
Storage Support
The development of SAN is booming in China. In fact, to support the transmission of various SAN signals over MSTP, GFP itself provides a transparent mapping user data encapsulation format (GFP-T). Implementing storage network interconnection on 10G MSTP has always been a hot topic. Some manufacturers promise to provide ESCON/FICON/FC interfaces, allowing storage switches to directly connect to MSTP for transparent service transmission. However, whether to use MSTP for SAN interconnection is highly debated in the industry. Some manufacturers believe that only WDM can provide fully transparent transmission, and when SCSI, FC, and SDH frames are mapped to each other, a converter must inevitably be added; moreover, the interworking of MSTP networks is far less convenient than open WDM, making it difficult for proprietary storage protocols of different companies to interoperate.
Intelligent Control
The ultimate goal of ASON is to achieve distributed network management capability, requiring network elements to have distributed intelligence, scalability, and collaborative working capability. On the control plane, an Overlay-based ASON model is implemented by defining professional interfaces such as UNI and NNI.
Since 10G MSTP is also located at the network core layer and plays a similar role in many aspects to the current ASON-supported DXC, it is certain that intelligence will first be provided on 10G MSTP. From the current situation, it is difficult to overlay a control plane on 10G MSTP in the short term, so the current focus should be on the interoperability between the DXC control plane and the 10G MSTP network management system.
On the basis of expanding the cross-connect matrix, 10G MSTP also needs to support mesh-network-based protection in addition to the original ring protection, thereby creating conditions for multi-route bypass protection. Some manufacturers claim that their new 10G equipment can have intelligent optical network functions in the future simply by inserting an intelligent card, but this claim requires more experiments to verify.