Next Generation SONET - An Analysis

Transcription

1 WHITE PAPER Author: Ramnarayan S SONET has been the predominant choice amongst carrier networks to carry their traffic. The term Next Generation SONET is typically used to denote the evolution of SONET to cater to the current needs of carrier networks. This paper aims to analyze the evolution, drivers, fitment in service provider networks and the technologies involved and standards that are involved in Next Generation SONET. A brief on probable alternatives is also discussed. Wipro Technologies Innovative Solutions, Quality Leadership

2 Table of Contents INTRODUCTION... 3 STRANDED BANDWIDTH... 4 VIRTUAL TRIBUTARIES/ARBITRARY CONCATENATION... 5 ISSUES INVOLVED... 6 VIRTUAL CONCATENATION... 6 LCAS... 8 GFP ALTERNATIVES CONCLUSION REFERENCES GLOSSARY ABOUT THE AUTHOR ABOUT WIPRO TECHNOLOGIES WIPRO TELECOM AND INTERNETWORKING Table Page of : Contents

3 Introduction SONET and its international counterpart, SDH, were conceived in the 1980 s as a platform for vendor inter-operability in optical networks. SONET has been the carrier s protocol of choice since then for its robust features and management capabilities. SONET transmission rates have also been steadily increasing with OC-768 (40Gbps) being in trials. The protocol was developed primarily to transport voice traffic as fixed rate timeslots through the network, thereby leaving it inefficient for carrying bursts of traffic that are characteristic of data networks. In the 1990 s, the paradigm shifted towards data centric networks and the Internet explosion took place leading to huge volumes of data being transferred over the existing infrastructure. SONET has tried to cope up with the data bandwidth explosion within its constraints, but only at the cost of huge bandwidth wastage due to its rigid hierarchy. STS-1 is the basic rate at which SONET operates. This provides for Mbps bandwidth. All other rates that SONET provides are multiples of STS-1. SONET by default provides for containers to fit a different rate of data other than multiples of STS-1 with a concept called concatenation where contiguous STS-1 frames are concatenated together to form a higher level signal. If the bandwidth required were 150 Mbps, then three STS-1s can be concatenated to form a STS-3c signal which can accommodate the bandwidth. However this scheme is inflexible due to the main reason that the signal rates in SONET have a fixed granularity and jumps by a factor of 4 leaving no intermediate rates. The following table lists the standard rates as defined by SONET. SONET SIGNAL BANDWIDTH STS Mbps OC Mbps OC Mbps OC Gbps OC Gbps OC Gbps Table 1: SONET Rates Wipro Technologies Page : 03 of 16

4 For example, the next rate after STS-1 (51.45 Mbps) is OC-3 (155 Mbps) and after that it is OC-12 (622 Mbps) and after that it is OC-48 (2.5 Gbps). If the requirement is less than any of the standard available rates, then it can be provided for only with a larger container with the difference in bandwidth going unused. The following table gives the packing efficiency of different client signals into SONET. The concatenation is assumed to be contiguous concatenation where successive STS timeslots are banded together to form a larger size container. CLIENT SIGNAL SONET FIT EFFICIENCY 10 Mbps Ethernet 100 Mbps Ethernet STS-1 (51 Mbps) 19% OC-3 (155 Mbps) 65% 1 GbE OC-48 (2.5 Gbps) 40% 10 GbE OC-192 (10 Gbps) 100% ESCON OC-12 (622 Mbps) 33% Table 2: SONET - Client signal mapping efficiency Stranded Bandwidth There was another issue that warranted attention, stranded bandwidth. Assume that a service provider services 4 different places A, B, C and D. The path from A to D can be either through B or through C. The provider has allocated STS timeslots between A and B and B and D and also A and C and C and D. NODE B 2 timeslots free 2 timeslots free NODE A NODE D 2 timeslots free NODE C 2 timeslots free Provision STS-3c between A and D!!! Figure 1: Standard bandwidth Page : 04 of 16

5 There remain 2 free STS-1 timeslots each on the path A-B and B-D and A-C and C-D. Now if there is a new request for a bandwidth of rate STS-3c, there is no way the provider can act on the request though there is bandwidth available in the network. Grooming may be considered if the network supported it, but there is no guarantee that the STS-1s reaching the destination through different paths arrive exactly in phase. The service provider requirement could be summed up as follows. To optimize network usage, there should be a flexibility of allocating right-sized pipes for carrying the client signals and not over-allocate bandwidth. Virtual Tributaries / Arbitrary Concatenation SONET provides for access to the lower level signals in an STS-1 frame by framing the signal in such a way as to provide for virtual tributaries. The VT level signal can be extracted from the stream without having to demultiplex the entire signal. The minimum container size in the VT hierarchy is the VT1.5 which was the bandwidth equivalent of Mbps or a T1. VT2, VT3 and VT6 are the other different levels of VTs in a STS frame. There can be a mix of different types of VT in a STS-1 frame. VT level grooming can also be done if the hardware supports it. Arbitrary concatenation is very similar to contiguous concatenation, but the difference is that a non-standard concatenation multiplier can be used and non-contiguous timeslots in a line segment can be used to form a virtual concatenated channel. However this scheme is proprietary to the vendors implementing this type of concatenation and is not widely used for inter-operability concerns. Allocating bandwidth at a VT granularity in increments is the best way to achieve optimal bandwidth utilization. The network element should be capable of handling virtual tributary signals. If the bandwidth utilization as noted in Table 2 were realized using VT mappings or arbitrary concatenation, then the bandwidth utilization may read as follows. CLIENT SIGNAL 10 Mbps Ethernet 100 Mbps Ethernet SONET FIT EFFICIENCY 7 x VT1.5 90% 2 x STS-1 99% 1 GbE 7 x STS-3c 95% 10 GbE 2 x STS-1 100% Table 3: SONET - Client signal mapping efficiency with VTs/arbitrary concatenation Page : 05 of 16

6 Issues Involved The issues that typically have to be looked at are summarized below. How to use the stranded bandwidth in the network without upgrading any of the existing legacy SONET ADMs/cross-connects in the network? What is the best way to carry data traffic over SONET? How to dynamically and efficiently resize the allocated bandwidth without loss of traffic? How to make sure that the loss of a link in a group doesn t fail the whole group? How to support multiple traffic types across the same physical infrastructure that can work across Layer 1 and Layer 2 traffic with a solution that scales rather than having a fragmented solution with a multitude of options? To address these issues, we need to look at the concepts of virtual concatenation, LCAS and GFP, that form the ingredients of next generation SONET. The rest of the paper discusses each of these technologies. One of the major drivers in the evolution towards data orientation of SONET is the increasing prevalence of Ethernet being extended into the Metro and WAN space. According to a study, almost 90% of internet traffic begin and end their lives as Ethernet frames. And by nature, this traffic is bursty and the existing rigid hierarchy SONET network is not able to carry Ethernet traffic efficiently. All the technologies that are being looked at below came from the problem of efficient transportation of Ethernet / packets across the existing infrastructure. Virtual Concatenation What is it? Virtual Concatenation is based on the principle that a number of small containers (e.g. VT1.5/ VT2/STS-1) can be concatenated and assembled to form a larger sized container. The containers (typically STS-1s) are then routed diversely across the network towards the same destination where the individual STS-1s are re-aligned and sorted to form the original payload. Virtual Concatenation has been defined for the following containers, VT1.5, VT2, VT3, VT6, STS-1 SPE and STS-3c SPE. A virtually concatenated group is referred to as <Base Container>-<n>v where n is a number indicating the number of containers of the base container type. For example, VT1.5-7v indicates that the group consists of 7 VT1.5s that are virtually concatenated, STS-1-4v indicates 4 STS-1s that are in a virtually concatenated group, STS-3c-7v indicates 7 STS-3cs in the group. There is also another school of representation which indicates the STS-1 count directly as in STS-21v, STS-7v etc. The former method is more standardized than the latter. The processing of the initial payload into the different channels is done only at the end points in the network. i.e. the payload is groomed into the different virtual containers at the ingress into the network and re-assembled and aligned at the exit for the network. Page : 06 of 16

7 How is it achieved? The basic premise in virtual concatenation is that the split-up original payload goes towards the destination as individual STS-1s which is then re-assembled and sorted. There should be no requirements on existing NEs that transit VC-ns part of a Virtual Concatenation Group (VCG or VC-n-Xv). Also there should be no strict routing constraints for operators. There are two different types of virtual concatenation, higher order virtual concatenation (HOVC) and lower order virtual concatenation (LOVC). The difference is in the base channel rate. If the channel rate is STS-1 and greater for a channel in a group, then the term HOVC is used. If it is sub-sts-1, then LOVC is used. Additional information would need to be passed along with the signal to indicate the order of the channel in the original payload superframe. The virtual concatenation overhead consists of the Multi Frame Indicator and the Sequence indicator. The Multi Frame Indicator is used to determine the differential delay and re-align the data to re-construct the original payload. The sequence indicator is an identifier assigned to each channel (VC-n) in the virtually concatenated group at the source to be used for reconstruction at the receiving end. In HOVC, the H4 byte of the path overhead is used to send VC related information. In LOVC, the VT path overhead byte V5 is used to send VC related information. In HOVC, H4 bytes from a 16 frame sequence make up a message for VC. Out of the 16 frames, VC uses 4, 5 are reserved and 7 are used by LCAS. In LOVC, V5/Z7 bytes from a 32 frame sequence make up a VC related message. STS-1s of the same VCG start with the same MFI SONET NETWORK (STS-1s may be routed diversely towards the same destination) STS-1s of the same VCG can arrive out of phase due to network transit delays Need for buffering!!! Figure 2: Illustration of differential delay for different channels of the VCG Page : 07 of 16

8 The MFI indicator is a running frame number, which is incremented with each new frame. In the above figure, the three STS-1s shown all have the same MFI. The next three STS-1s transmitted will have their MFIs incremented by one. At the destination, due to different delays, the MFIs will no longer necessarily be the same for the three STS-1s. For example, the first STS-1 may have a trace that is 500 ìs (four frames) faster than the second and third STS-1. Hence at any given instance, the MFI number of the first STS-1 at the destination will be four higher than the other STS-1s. In order to extract the original frame, the destination node needs to compensate for the different network delays by delaying the first STS-1 by 500 ìs. The 12-bit MFI number allows end nodes to compensate for up to 256 ms of differential delay. The compensation is typically done by buffering the frames arriving out of phase and before re-ordering. The source node labels each STS-1s, in a virtually concatenated channel, with a sequence number indicating its relative position. An STS-Xv channel will have SQ number zero to (x-1). In Figure 2, the upper STS-1 is assigned SQ #0, the middle SQ #1 and the lower SQ #2. At the destination, the STS-1s are reordered according to the sequence numbers to guarantee that the frame is extracted and collated in order. This SQ number relieves the network management of having to keep track of the order of each individual trace through the network. As long as the intended STS-1s are routed to the destination node, the order within a channel is sorted out at the destination. Advantages Flexibility of provisioning right sized channels across the SONET/SDH network. The individual virtually concatenated channels need not be on the same path. The channels are not constrained to the same transport channel also. E.g. STS-1, STS-12 etc Concatenation facility is only required at the termination point. Choice of bandwidth granularity ranges from VT1.5s to STS-3c. Provides additional transport sizes in addition to the standard SONET hierarchy. LCAS What is it? Virtual Concatenation provides an avenue for channeling a client signal across different containers. If any VC-n in a VCG fails, then the whole frame is deemed to have failed. Typically in the case of packet framing, it is not needed to discard the entire frame. The bandwidth could be considered to have decreased. Also there is need for flexibility in increasing / decreasing the bandwidth dynamically based on different considerations as time of day. Today s networks don t offer too much of flexibility on that score. LCAS (Link Capacity Adjustment Scheme) was designed to enhance Virtual Concatenation by adding the capacity to dynamically alter SONET bandwidth without any traffic hits through a signaling mechanism. It also provides a capability for temporarily removing failed member links from a Virtual Concatenation Group. LCAS assumes that in cases of capacity initiation, increase or decrease, the construction or destruction of the end-to-end path of each individual member is the responsibility of the Network and Element Management Systems. Synchronization of changes in the capacity of the transmitter and the receiver shall be achieved by a control packet. Each control packet describes the state of the link during the next control packet. Changes are sent in advance, so that the receiver can switch to the new configuration as soon as it arrives. Page : 08 of 16

9 If the different members of a Virtually Concatenated channel are diversely routed, LCAS can provide a fault recovery mechanism. The failed member(s) are flagged as failed by the LCAS sink. The source then marks the member(s) as do not use and reduces the mapped information to fit the now-available bandwidth. The result is a recovery that keeps a connection alive in the non-failed member(s) at a reduced data rate. The NMS also has the option of permanently deleting the failed member(s). How is it achieved? Just as in the case of Virtual Concatenation, the protocol byte communication happens though the H4 or V5/Z7 overhead bytes. For High Order VCs, it is communicated in a Control Packet carried in bits 1-4 of the H4 byte carried across a 16 frame multiframe. For Low Order VCs, it is communicated in a Control Packet in bit 2 of the Z7 byte Carried across a 32 frame multiframe. The term multiframe implies that the final protocol message can be reconstructed only after getting the n frames. i.e. for a 16 frame multiframe, the protocol is carried across in 16 successive frames. The entity looking at the same can make a distinction only after receiving the 16 frames. LCAS is a two-way handshake protocol. Status messages are continuously exchanged and consequent actions taken. Each STS-1 carries one of six LCAS control commands. Fixed Add Norm EOS Idle Do not use LCAS not supported on this STS-1. Request to add this STS-1 to a channel, thereby increasing the bandwidth of an existing channel or creating a new channel. This STS-1 is in use. (End of Sequence) This STS-1 is in use and is the last STS- 1 of this channel, i.e. the STS-1 with the highest SQ number. This STS-1 is not part of a channel. This STS-1 is supposed to be part of a channel, but is removed due to a broken link reported by the destination. Summarizing, in the forward direction from Source to Sink the following commands flow Multi Frame Indicator (MFI) (part of Virtual Concatenation) Sequence Indicator (SQ) (part of Virtual Concatenation) Control (CTRL) - IDLE, ADD, NORM, EOS, DNU, FIXED Group Identification (GID) which is used for identification of the VCG. All members in the same VCG have the same value in all the frames with the same MFI In the return direction, Sink to Source Member Status (MST) containing the status of all the members in the VCG Re-Sequence Acknowledge (RS-Ack) to indicate back to the transmitter that the changes have been accepted. For both directions Cyclic Redundancy Check (CRC) over the control packet Page : 09 of 16

10 Advantages Offers the flexibility of dynamically resizing bandwidth without bringing down the entire VCG or affecting service. LCAS provides load sharing capabilities where if a member of a VCG fails, the VCG is still up, but operates with reduced bandwidth. Does not impose any restriction on the intermediate NEs. Only the source and sink NEs need to have this capability. LCAS defined for use over the Optical Transport Network (G.709 OTN) also. Protocol agnostic. GFP What is it? GFP (Generic Framing Procedure) provides a generic mechanism to adapt traffic from higher-layer client signals over an octet synchronous transport network. Client signals may be PDU-oriented (such as IP/PPP or Ethernet MAC), block-code oriented (such as Fibre Channel or ESCON), or a constant bit rate stream. It has better performance than Packet Over SONET (POS) framing and agnostic to L2 and above. It can also be applied to dark fiber. It can be considered as an alternative transport mechanism to ATM and HDLC framing. The figure below shows a fit of GFP at different layers in the networking domain. e.g. IP (PPP) ATM HDLC GFP 12 Ethernet ATM SONET/SDH Pt to Pt Ring GFP RPR/GFP PHY DWDM Layer Figure 3: Generic Framing Procedure for efficient packet transport GFP has a low overhead as compared to HDLC like framing (typically used in POS). There is no inflation factor as the payload size increases. GFP is also rate matched to Ethernet. There are two forms of GFP, i.e. Frame-mapped GFP and Transparent GFP. The former is typically used with variable payload length frames (such as IP packets/ethernet MAC frame) and the latter for constant bit rate block coded data signals (STS-1 frame) Also transparent GFP is used in a point to point topology whereas frame-mapped GFP can be used in point to point, resilient packet ring, aggregator topologies. Page : 10 of 16

11 A look at GFP internals GFP consists of a core header and a payload. The payload consists of a payload header, a Frame Check Sequence and the payload itself. Also there are client specific aspects (payload dependent) and common aspects (payload independent) of GFP. The GFP frames can be broadly classified as follows. Client Frames Client Data Frames Client Management Frames Control Frames Idle Frames Core Header 4 Octet Transmission Order Payload Area Ocets Bits Bit Transmission Order Figure 4: Frame Format for GFP user frames The figure above provides an understanding of the frame layout in GFP The core header is intended to support frame delineation procedures and essential data link operations functions independent of the higher layer PDUs. The GFP Core Header consists of a PDU Length Indicator field and a Core Header Error Check field. Page : 11 of 16

12 Octet 1 Payload Length Indicator (MSB) Octet 2 Payload Length Indicator (LSB) Octet 3 core Header Error Control (MSB) Octet 4 core Header Error Control (LSB) Figure 5: GFP Core Header formal The payload length indicator field indicates the size of the payload and the core Header Error Control field has the CRC-16 checksum of the header bytes with single/multiple bit error correction enabled. The payload field can be looked at as follows. Payload Header Payload Payload FCS (optional) Figure 6: GFP Payload The payload header typically consists of a 4 octet header consisting of the type and two octets for the type Header error control and with extension headers can be up to 64 octets in length. The Type field consists of a 3-bit Payload Type Identifier (PTI), a 1-bit Payload FCS Indicator (PFI), a 4-bit Extension Header Identifier (EXI) and an 8-bit User Payload Identifier (UPI). The payload may be up to X octets where X is the size of the payload header. The payload FCS if present forms the last 4 octets of the payload and can be identified by the FCS indicator bit in the payload header. Advantages Enables transport services for Layer 1 or Layer 2 payloads (IP, PPP, Ethernet, HDLC at Layer 2, Fibre channel, FICON, ESCON etc at Later 1) The mapping is uniform across all Path types. The mapping is also uniform across all Ethernet types. This minimizes cost by maximizing equipment commonality. Highly scalable. From 10 Mbps to 10 Gbps and beyond. All of the relevant MAC layer information, from Destination address through Frame Check Sequence (FCS) inclusive, is preserved intact by the mapping. This maintains a clear distinction between layers. Since the FCS is preserved, the native Ethernet error detection capability is protected. Consequently, the error detection capability is not degraded. The mapping doesn t inflate the frame length in a non-deterministic way as what HDLC does and hence the throughput capacity is predictable. This eases network planning and ensures that the throughput is independent of data content. The throughput is maintained at a high rate by use of a robust delineation mechanism and by the deterministic, non-inflationary encapsulation. Delineation is done by two consecutive chec field matches vs. computed chec. Page : 12 of 16

13 Alternatives G.709 or the digital wrapper for the OTN can be considered an extension of SONET. It is a framing/encapsulation technique for DWDM networks. It is used to monitor and manage DWDM wavelengths in optical transmissions. A section of digital overhead bandwidth is added to client bandwidth before it is sent over the transport network. It has facilities for performance monitoring, fault detection and protection switching. Error correction is based on Forward Error Correction techniques. It is compatible with all existing network protocols because it operates separately from the client signal being transported. There are equipment vendors who use sub-rate multiplexing over DWDM using thin muxes to get better efficiencies over SONET. The argument is that VC is more tailored towards a single channel and that robs the network of its efficiency. GFP has its alternatives in the X.86 framing format (also called as LAPS, Link Access Procedure for SDH). It can be defined as a type of HDLC that includes data link service and it is the protocol used in transporting IP packets over SDH networks. It has been standardized by ITU in the recommendation X.85/Y.1321 (IP over SDH over LAPS) It is also fully compatible with RFC 2615 (PPP over SONET) GFP is better than LAPS because there is no inflation factor and maintains a fixed overhead which is almost equal to the minimum overhead of LAPS. Also the chec field in GFP allows for single bit error correction whereas a single bit error in LAPS can cause mis-alignment. Also GFP allows multiple protocols from different ports or links to share the same transport path resulting in more efficient use of available bandwidth. Also GFP supports RPR operation through the use of extension headers. Conclusion New technologies that are extensions to SONET make it more palatable to carrying data, viz. virtual concatenation, LCAS and GFP. Though it is premature to predict the demise of SONET and the appearance of better technologies, the evolution of SONET in trying to take care of deficiencies has allowed it to be in the race. Also the carrier networks which has been used to the five nines reliability offered by SONET will not migrate easily to other technologies unless they offer something well and far above what SONET provides. References 1. Telcordia, SONET transport systems: common criteria, GR-253-CORE Issue 2, January ITU, Link Capacity Adjustment Scheme, G.7042/Y.1305, November ITU, Generic Framing Procedure, G.7041/Y.1303, November ITU, G.707 Network Node Interface for the Synchronous Digital Hierarchy (SDH), April Information/white-papers from the web-sites of the following organizations. AMCC, Agere, Agilent, Cisco, Lucent, OptiX, PMC-Sierra, ATIS-T1X1.5, Xilinx Page : 13 of 16

14 Glossary ADM ATM DWDM FCS GFP HDLC HOVC LAPS LCAS LOVC M(L)SB MFI Add Drop Multiplexer. An equipment in a SONET network that can add and drop STS/VT rate signals into a SONET line. Asynchronous Transfer Mode Dense Wavelength Division Multiplexing Frame Check Sequence Generic Framing Procedure. Frame format defined to efficiently carry data or constant bit stream data across carrier networks. High level Data Link Control Higher Order Virtual Concatenation. Virtual Concatenation on STS-1 rate signals and above Link Access Procedure for SDH Link Capacity Adjustment Scheme. A protocol defined on top of virtual concatenation to dynamically resize bandwidth of the connection. Lower Order Virtual Concatenation. Virtual Concatenation on sub STS-1 rate signals. E.g. at VT rates Most (Least) Significant Byte Multi-Frame Indicator. It is a byte used in the path overhead to indicate phase of V1 to V4 bytes in a VT superframe OC-n Optical carrier at rate N, N=1, 3, 12, 48, 192, 768 OTN PDU POS PPP SDH SONET STS VCG VT Optical Transport Network Protocol Data Unit Packet over SONET Point to Point protocol Synchronous Digital Hierarchy. An international standard similar to SONET. SONET can be considered as a subset of SDH. Synchronous Optical Network. Transport protocol used in North American carrier networks. Synchronous Transport Signal, the basic electrical unit of transmission in SONET Virtual Concatenation Group. A term used to denote all the STS-1s/VTs that form the bandwidth. Virtual Tributaries, a concept in SONET allowing access to lower rate signals in a STS signal. Page : 14 of 16

15 About the Author Ramnarayan S is a Technical Consultant with Wipro Technologies in The Optical Networks domain. He was a design engineer and worked with various technologies across different domains ranging from voice switching, routers, network management and optical ADMs. About Wipro Technologies Wipro is the first PCMM Level 5 and SEI CMMi Level 5 certified IT Services Company globally. Wipro provides comprehensive IT solutions and services (including systems integration, IS outsourcing, package implementation, software application development and maintenance) and Research & Development services (hardware and software design, development and implementation) to corporations globally. Wipro s unique value proposition is further delivered through our pioneering Offshore Outsourcing Model and stringent Quality Processes of SEI and Six Sigma. Page : 15 of 16

16 Wipro Telecom and Internetworking Wipro Telecom & Internetworking, a division of Wipro, offers comprehensive solutions for telecommunication to confront challenges, and convert every challenge into an opportunity. With over two decades of telecom experience, Wipro T&I offers a wide range of solutions various domains such as wireless networking, broadband (data, optical and access networking), voice switching, network management hardware design etc. Wipro T&I also provides several IPs, components and reference solutions in these domains which help our customers in saving costs while providing the time-to-market advantage. We offer complete consulting, architecting, design, implementation and maintenance services to the telecom equipment manufacturers. Wipro Technologies can be reached on the Web via the URL Worldwide HQ Wipro Technologies, Sarjapur Road, Bangalore , India. Tel: U.S.A. U.K. France Wipro Technologies Wipro Technologies Wipro Technologies 1300, Crittenden Lane, 137 Euston Road, 91 Rue Du Faubourg, Mountain View, CA London, NW1 2 AA. Saint Honoré, Paris. Tel: (650) Tel: +44 (20) Tel: + 33 (01) Germany Japan U.A.E. Wipro Technologies Wipro Technologies Wipro Limited Am Wehr 5, # 911A, Landmark Tower, Office No. 124, Oberliederbach, Minatomirai 2-chome, Building 1, First Floor, Frankfurt Nishi-ku, Yokohama Dubai Internet City, Tel: +49 (69) Tel: +81 (04) P.O. Box , Dubai. Tel: +97 (14) info@wipro.com Wipro Technologies Innovative Solutions, Quality Leadership Page : 16 of 16