Testing Ethernet over SONET/SDH
By GUYLAIN BARLOW, Innocor Ltd. -- New standards provide new capabilities -- and add new test challenges as well.
Data traffic levels continue to increase. As a result, there is a desire to optimize transport networks originally designed for voice to accommodate this traffic. The majority of metro and long-haul networks are based on SONET and SDH, as these protocols carry fixed-bandwidth circuit-switched voice very effectively. Ethernet over SONET/SDH (EoS) optimizes transport networks to adapt to data and storage traffic, particularly for Ethernet and Fibre Channel applications. This article will discuss the basic functions and test requirements of virtual concatenation (VCAT), which provides flexible bandwidth groupings for SONET/SDH; Link Capacity Adjustment Scheme (LCAS), which provides dynamic bandwidth settings; and Generic Framing Procedure (GFP), which provides a protocol-agnostic frame container.
The environment of EoS is one of convergence. The traditional approach in both the vendor and service provider environments has been a partition of knowledge between transport technology and Ethernet technology. One of the fundamental differences is that the transport environment is dominated by static, synchronous connections while data networks carry bursty, less predictable traffic. As a consequence, the monitoring of a transport network has involved verifying the status of alarms and synchronous errors. Conversely, data network monitoring involves counting frames, throughput, and how fast frames are delivered.
Equipment capable of carrying Ethernet frames over SONET/SDH equipment has been available for some time. However, the mechanisms to provide such a function have been proprietary -- mainly due to a lack of standards prior to VCAT, LCAS, and GFP -- which prevented interoperability among vendors and the development of third-party test equipment. This also limited the capability to broadly deploy the service.
Definitions and requirements
VCAT, an extension of SONET/SDH (ANSI T1.105/ITU-T G.707), maintains the required SONET/SDH influence within the context of service reliability. On the other hand, the extension of VCAT with LCAS (ITU-T G.7041) introduces a dynamic element to SONET/SDH that is more akin to a data protocol environment. To complete the picture, GFP (ITU-T G.7042) either provides a one-to-one mapping between GFP frames and Ethernet Protocol Data Units (PDUs) in frame mode (GFP-F) or fixed-length blocks in transparent mode (GFP-T). GFP-F is effectively a data protocol where information monitoring must be based on frame units. Since EoS encompasses VCAT, LCAS, and GFP, one of the requirements is to examine the needs in testing both the transport and data aspects of the network.
Testing SONET/SDH includes the injection and monitoring of alarms at the line/regenerator section, section/multiplex section, and path levels. VCAT acts at the path level with the introduction of member links in a virtual concatenation group (VCG). Such members are identified and managed via a sequence number, unique to each member, and a multiframe indicator, which increments sequentially as frames are transmitted. At least three defects can be validated in VCAT: Loss of Multiframe (LOM), based on the multiframe indicator; Loss of Sequence (SQM), based on the sequence number; and Loss of Alignment (LOA), when the maximum differential delay is exceeded. Differential delay is the relative arrival time measurement between members of a VCG.
VCAT is provided at the edge of the network where individual members are transported transparently in the core network. Since each member may use a different physical route through a network, VCAT supports the buffering of the information to account for delays. When buffers are exceeded, an LOA situation arises. In testing differential delay, it is necessary to be able to inject and measure the differential delay up to the maximum supported delay on the equipment. The absolute maximum is 256 msec.
Another aspect of VCAT is the size of each member and the maximum number of members in a VCG. Since the main premise of EoS is to carry Ethernet, enough bandwidth is aggregated to correspond to 10 Mbits/sec, 100 Mbits/sec, or 1 Gbit/sec for Gigabit Ethernet. The SONET/SDH containers, which members use in VCAT, are either high-order or low-order paths. Supported high-order paths in SONET are either STS-3c or STS-1, while in SDH they are either VC-4 or VC-3, where the latter is typically mapped via TUG-3. In low-order, they are generally VT1.5 in SONET and VC-12 in SDH. Figure 1 shows the containers and sample applications.
LCAS provides signaling functions to dynamically add and remove members from a VCG. The test functions at this level verify the different states of the signaling protocol including the timing of control messages. Figure 2 illustrates the use of the H4 byte in the path overhead for high-order mappings. Figure 3 shows the use of the Z7/K4 (SONET/SDH) byte for low-order. Test functions to provide error injection and monitoring associated with LCAS messages are also required. This includes CRC errors and unexpected control messages.
In GFP, more specifically GFP-F, detailed testing is inspired from data protocols. With GFP-F, data is carried asynchronously, where idle frames are transmitted when no traffic data is available. In this context, the concepts of traffic bandwidth, bandwidth profiles, and traffic streams apply. Traffic monitoring becomes a matter of ensuring that frames are not lost or received out of order. At the service level, measuring latency as in IETF RFC 2544 is another important test capability to ensure service quality. The measurement, although typically accomplished directly at the Ethernet interface, may also be run over GFP on the EoS interface.
EoS is helping service providers deliver end-to-end Ethernet and storage services. Its primary importance is to provide flexibility and more efficiency in carrying data services on the SONET/SDH infrastructure. The combination of VCAT, for better bandwidth granularity, LCAS, for dynamic bandwidth, and GFP, as a generic framing mechanism, extend the transport network capabilities. This introduces a new set of requirements to test and validate the service. One of the impacts is the convergence of the transport centric philosophy based on high reliability and data network efficiency particularly in the sharing and reuse of bandwidth. The demands in test execution will be inspired by both viewpoints, and this can only improve networks and the expertise of those who build them.