For transporting voice and data, synchronous optical networking (SONET) has been the one ubiquitous, primary network architecture since the mid-1980s. Sure other protocols are moving in, such as asynchronous transfer mode (ATM) and Internet protocol (IP), but folks in the know say that SONET is holding its own.
"SONET was designed to handle network management activity within the SONET stream," begins Marek Tlalka, product line manager for SONET devices at Conexant Systems Inc. (San Diego, CA). "There is no requirement for an overlay network anymore to maintain SONET." That differs from previous legacy Plesiochronous Digital Hierarchy (PHD) networks, "where companies had to put in an overlay network on top of a transport network to collect all the maintenance information and to do the network configuration from a central operations systems center," Tlalka says.
Increasingly, however, unprecedented demand is pressing SONET, a tried and true standard. "As the Internet began to pervade the landscape, the percentage of voice traffic relative to the total traffic has shrunk from 80 percent to 20 percent. At the same time, data traffic has grown from 20 percent to 80 percent," notes Bill McNeill, product marketing engineer at Richardson, Texas-based Anritsu.
What's pressing food chain contenders, those challenged by the service requirements of today's IP based customers, is that "legacy SONET networks were designed to provide circuits for voice transport. The bandwidth needs of the legacy end users were predictable and slow to change." According to McNeill, "In contrast, bandwidth needs of today's data users are unpredictable and scale rapidly."
The Difference Between SONET and SDH
The SONET and SDH protocols are almost identical. According to James Frodsham, vice president of marketing at Qtera (Boca Raton, FL), "The SONET multiplexing hierarchy starts at OC-1, or 51.8 Mb/s, and moves up in increments of this base rate; (meaning that) OC-3 = 3 X OC-1 or 155 Mb/s and so on. In SDH the base rate starts at 155 Mb/s and this rate is referred to as STM-1. At STM-1 and above, the SONET and SDH speeds are the same." For example, OC-12 = STM-4 = 622 Mb/s, and OC-48 = STM-16 = 2.488 Gb/s, he says.
Frodsham explains, "The difference in the base rate is driven from legacy electrical services. In North America, SONET was required to support the carriage of DS3, or 45 Mb/s services. In Europe, SDH was required to support the carriage of E3, or 140 Mb/s services. The OC-1 and STM-1 rate definition was a compromise which enabled these services to be managed efficiently while allowing for the rate convergence at 155 Mb/s."
"In today's network, these legacy issues are no longer significant." Adds Frodsham, "The SONET/SDH protocol provides a unified and globally standardized multiplexing hierarchy, which enables the efficient transport of any service type anywhere in the world."
Advantages of SONET
Entrenched and refined, SONET has several distinct advantages over the contenders, such as guaranteed level of circuit availability because of SONET's traffic protection scheme. According to Bob Chomycz, a SONET/fiberoptic systems engineering consultant and CEO of Telecom Engineering Inc. (Thunder Bay, Canada), as well as the author of several notable books on SONET, "If there is a fiber cut, the SONET system will switch traffic in 50 milliseconds or less onto alternate fibers. As a physical layer protocol it works quite well."
Chomycz cites another SONET advantage: "Channels can be tested and monitored to a certain level of quality of service. With SONET you know if you deploy an OC-48 system you can test each one of the OC-48 channels and be assured that they meet a certain level of quality using an industry recognized standard testing method. With IP it's unclear how you actually test an IP path, at is full data rate, to a certain quality of service.
In addition, various network configuration activity can be carded on the data communication channel from SONET Because of its duplex environment, "there is really no risk that you will not be able to get to a network element to run maintenance activities. You always have a redundant path," notes Tlalka.
Quality of Service Guarantees
The ability to provide their customers with an array of quality of service guarantees (QoS) is clearly a reason why every major carrier and telco has SONET in its network. For example, when a carrier addresses the performance of their transport infrastructure they can "specify a know-bit-error-rate performance for the signal, and they (can) specify an availability requirement," explains Frodsham.
"The transport network is a connection-oriented system, and, regardless of the transport distance, we have to guarantee the error rate performance of the signal across the facility," says Frodsham. "SONET provides a built-in means of providing these assurances."
Availability is also a key measure. Adds Frodsham, "That's why we have different protection mechanisms intrinsic to the transport equipment. If there is an equipment failure, or there is a fiber cut, we can guarantee the transmission network will automatically restore itself to such an extent that, from a client perspective, the failure is completely transparent."
"For private line TDM services, QoS was a transport problem and the transport network was designed to provide guaranteed rates and availabilities." Frodsham notes, "Moving forward in terms of ATM services, or IP VPN (Internet protocol virtual private network) services, the carriers are relying upon statistical multiplexing to increase the bandwidth efficiencies of the physical layer. It sort of increases the complexity of the problem."
In addition, Frodsham contends, "No matter how many other users are sharing that infrastructure, the carriers need to be able to guarantee a minimum throughput speed and network availability for each and every end-user."
Emerging Trends for SONET Network Management
Monitoring optical wavelengths in a fiber is definitely a clear network management trend. Chomycz says that with the emergence of WDMs (wavelength division multiplexers) and multi-wavelength SONET systems, optical transmissions from numerous SONET terminals can now be placed onto one optical fiber. "Therefore, there is a need to be able to monitor each SONET terminal's optical wavelength in a fiber," says Chomycz.
Prior to WDMs, says Chomycz, "you would monitor the total optical power in a fiber, knowing that only one wavelength existed in the fiber. Now, where WDMs are deployed, there is more than one wavelength in a fiber. Equipment that was used to monitor the total optical power in a fiber will not tell you how each individual wavelength in the fiber is performing."
According to Chomycz, monitoring the power level of each individual wavelength is required to insure that they are all present and equalized properly. "You could have, say, an eight wavelength system. If you are monitoring the total power you could have one wavelength missing and you may not notice it. Whereas if you are monitoring each individual wavelength, you would definitely know a wavelength was missing and see the levels of all the other wavelengths."
Real-estate gain due to reduced component and equipment size, DWDM compatible SONET systems, and numerous SONET systems working off one or two fibers, are other trends that Chomycz sees. Interest in the optical cross-connect (OXO) is another.
Management is one of the key challenges associated with bringing OXO technology into the network. Frodsham argues that while different implementation approaches are being considered, "we will still need to figure out how to performance monitor the optical cross connect itself and (to) know that the device itself is operating as specified. We need to know that we have connection verification and connection management capabilities so that the optical cross connect switches we were expecting to happen from a switching function did indeed happen. And we will need to verify not just that the connection through the optical cross connect was executed properly, but indeed that the connection through the network was managed properly."
What Silicon Brings to the Game
Tlalka contends that once you get to the box, a OXO or a multiplexer, or a router that connects directly to the ring, supporting management level software via silicon is a must. "The key is to provide in the silicon all of the features that are required by the standards -- whether it's the European standard or the American standard." Tlalka says that off-loading from the host processor functions such as error counting; alarm detection, activation and deactivation; and automating other network maintenance functionalities is ripe for silicon technology.
"Another trend in the SONET equipment is (that) more and more lines are being added to a system. A single processor is now responsible for managing a lot more input and output lines than in legacy systems. So, if that processor needed to go out and really do manual configurations of every port, it would quickly run out of real time." Suggests Tlalka, "The key is to provide as much information and automation as possible right in the silicon in the devices."
Silicon technologies can enable an automated response to network events and protection switching. "What was done in the past (was) a processor would get an interrupt, would have to detect and analyze what the source of the interrupt is, and then respond by setting another register," Tlalka says. According to him, these functions can now be done automatically in the silicon. "The processor basically finds out that the maintenance activity (has) just occurred, within milliseconds, but not immediately, and it can then send that information to the management systems."
Regarding protection switching, Tlalka adds, "there is a time requirement of 50 milliseconds to respond because the network is so fast that you lose a lot of data with every millisecond. That's another thing that's being automated in silicon right now so the hardware will actually make the decision to switch and the processor communicates that to the operations systems."
As for error counting, "It used to be that error counters were really small in the silicon, mainly to save silicon space because they require registers." Tlalka adds, "Now that we can put more functionality in the silicon because of smaller geometry, we're expanding error counter sizes such that a system really needs to read an error counter, at most, once a second and not more often than that."
Relieving the host processor of maintenance data load can also be achieved by putting a programmable state machine on a silicon device. Higher density of ports per system, and their respective maintenance activity, is spurting this need.
"What silicon is going to do, especially with programmable features showing up on the devices, is to allow for a single box in the same space to handle a lot of higher volume traffic and a lot lower channelization down to DS1. That's going to facilitate higher density circuits and lower level channelization." According to Tlalka, this facilitates the transition from TDM type circuits to high capacity packetized networks. "Because of all the legacy equipment right now, which supports channelization, there has got to be this transition period. And this channelization is what requires a lot of board space and a lot of processor power." Tlalka says, "High density silicon is facilitating that transition and allowing for it to happen."
Moving Forward
"SONET will continue to remain a viable framing protocol," in the words of McNeill. "However, some elements of the SONET protocol will move into the optical layer or service layer. SONET protocol elements that were artifacts of the voice-centric network will disappear from newer IP-centric equipment."
"There are other protocols around that are trying to muscle into SONET's arena -- ATM over fiber or IP over fiber, for example," says McNeill. But Chomycz adds that ATM or IP over fiber lack the standardization, the reliability, and the many desirable characteristics that SONET has today.
* Write In 703 or Reply Online
EDITORIAL EVALUATION
Write in Number on Reader Service Card I found this article:
Very Useful Useful Not Useful 704 705 706
