Shedding Light On Sonet
March 20, 2000
By Darrin Woods
Sonet is slowly moving into everyday business life as bandwidth requirements within the enterprise increase. Sonet, or synchronous optical network, is the standard for transmitting synchronous data on optical-electrical media. It allows the entire contents of a 650-MB CD-ROM to move from coast to coast in less than one second. Businesses that can't contain their entire workforce in a single building are adding Sonet rings to interconnect offices in MANs (metropolitan area networks), and packet-over-Sonet has the potential to supplant ATM in a local area network and across a wide-area network.
A Little History
Sonet was born out of necessity. As the population expanded and communication demands grew, copper--once the transmission material of choice--ceased to be economical or practical to carry the huge number of calls nationwide. Copper also was highly prone to electrical spikes from storms and other electrical interference. Enter fiber optics. Now data that once required hundreds of copper cables could be directed down a glass fiber only slightly thicker than a human hair. Carriers jumped on this technology and tried to one-up each other in the amount of fiber each had. To keep up with carriers' needs, vendors created complex systems to multiplex traffic onto these tiny strands. Unfortunately, these systems were usually proprietary, and interconnection of different products even from the same vendor was difficult. Interconnection between two carriers (each using different hardware vendors) was that much more difficult. Something had to be done.
In 1984, the Exchange Carriers Standards Association (ECSA) proposed a method to interconnect the fiber optic systems from multiple vendors. Bellcore extended the original ECSA idea in 1985 and proposed what we now know as Sonet. In 1988, the initial Sonet standards were approved as ANSI documents T1.105-1988, which described optical rates and data format, and T1.106-1988, which described the physical interface.
Sonet signals are referenced in two ways: STS (synchronous transport signal) is the electrical portion, and OC (optical carrier level) is the optical portion. Although Sonet was designed to eliminate the electrical transmission of data, STS is used for very short distances, usually only within a switch cabinet. Until pure optical switching is available, the electrical equivalent is necessary. STS-x describes frame generation within a switch, since it is done electrically; OC-x describes transmission of the signal from point to point. Because Sonet sends 8,000 STS frames per second--or one frame every 125 microseconds, the same frame rate that has been around since the DS-1 was invented--it's easy to incorporate current transmission timings.
Bandwidth ranges from 51.84 Mbps at the OC-1 level to 9953.28 Mbps at OC-192. There are specifications for higher bandwidths, with some vendors talking about OC-768 products--equivalent to seven CD-ROMs transmitted in one second--but these specifications have not been finalized. At the physical OC-x level, data travels in one of two ways--WDM (wave-division multiplexing) or DWDM (dense-wave-division multiplexing). WDM pulses a single laser to transmit data.
The faster this laser can be pulsed, the more bandwidth that can be pushed through the fiber. WDM can effectively pulse a laser at OC-48 speeds. To reach higher bandwidths, however, the size of the pipe must be increased. Enter DWDM, which achieves a higher bandwidth by combining multiple OC-48 WDM lasers (each operating at a different wavelength)--essentially using the same pipe but enlarging it by transmitting more wavelengths of light.
You have to take care when purchasing equipment. If you choose to buy lit fiber from a LEC (local exchange carrier), the carrier probably will specify and supply the equipment. Even though Sonet is a standard, there are subtle differences in the way some bytes are used in the overhead by different manufacturers. Therefore, it's important to match the LEC equipment.
Alternatively, dark fiber is available for purchase from LECs, and is used to connect local offices in a campuswide type of environment. You can usually choose your own equipment when you purchase dark fiber.
Building a Sonet Frame
When data enters a Sonet switch from an external source, it is placed into an STS frame. An STS-1 frame can be thought of as a 9x90, 8-bit-per-byte (octet) matrix (see "An STS-1 Frame," below). All bytes are transmitted in order, from the first bit of the first byte in the upper left corner to the last bit of the last byte in the lower right corner. It would be too confusing to display the data as a straight line of 810 bytes and show where everything is located within the frame, hence the matrix.
The STS frame is divided into two sections: the TOH (transport overhead) and the SPE (synchronous payload envelope), which are roughly analogous to the header and payload sections of a typical data frame. The TOH is subdivided into the SOH (section overhead) and LOH (line overhead).
The SOH occupies 3 bytes x 3 bytes of the TOH and works at the frame level. It's the lowest level just above the physical layer, and controls how the frame is transmitted from one regenerator to another.
The LOH occupies 6 bytes x 3 bytes of the TOH and handles the SPE's placement in the frame; it also tells the ADMs (add/drop multiplexers) how to extract the SPE.
The SPE, meanwhile, contains one column of POH (path overhead) and 86 columns of payload. The POH contains information relevant to the payload data and its position in the SPE.
The SPE's payload can vary, depending on the type of data in the frame. If one DS-3 is being mapped to the frame, it is placed directly into the SPE payload. But if sub DS-3 services are being transmitted, a new container must be defined. This new container is called the VT (virtual tributary) and is the only one for subrate services an SPE can carry. VTs are defined for four types of service: DS-1 (VT 1.5), E1 (VT 2), DS-1c (VT 3) and DS-2 (VT 6). Each type carries a different bandwidth and therefore occupies a different amount of space within the payload. VT 1.5 occupies 9 bytes x 3 bytes (three columns), VT 2 is four columns, VT 3 is six columns, and VT 6 is 12 columns.
It's difficult to mix and match VTs within a payload, but Sonet solves this problem by defining a VTG (VT group). There are seven VTGs within an STS-1 frame. Each VTG occupies 12 columns of the payload (see "Virtual Tributary Groups in an STS-1 Frame," above). VTs are matched and placed into a VTG that contains identical services. A single VTG can contain four VT 1.5s, three VT 2s, two VT 3s or one VT 6. A pointer is created to define what type of service any VTG is carrying. Before the seven VTGs are placed in the payload, they are byte-interleaved with two empty columns of 9 bytes each, located in columns 33 and 62. All this grouping lets Sonet sort out each channel, down to the DS-1, that's being transmitted from all the others.
Timing Is Everything
With the high rates Sonet can achieve, it is necessary to have some way to accommodate imperfections. Timing is the key to ensuring the payload gets into the frame and the frame is delivered when it is expected. It is inefficient to have empty frames while the switch waits for data or needs to adjust for timing issues. For this reason, Sonet was designed to have floating payloads. The entire SPE floats within a frame.
Two bytes in the LOH, H1 and H2 (see "Overhead Byte Breakdown," above, for a full diagram of bytes in the transport and path overheads), indicate where in the frame the SPE begins (see "Floating SPE," page 54). An SPE will therefore span two frames in most cases. This is not wasteful, because where one SPE ends, the next begins. It is possible for an SPE to be contained entirely within one frame, though it will actually be in the next one from which it is offset.
This level of floating continues down to the VTG and VT. An individual VT can float within its VTG, and the VTG can float within its SPE as specified by H4. As a result, Sonet can account for any timing difference on the networks where individual signals are on different timing sources.
Once all the data is encased in the SPE, error detection begins. An even parity byte is calculated for each of the SOH, LOH and POH, and placed in the following frame's B1, B2 and B3 byte location. The parity byte contained in each frame is actually for the information contained in the previous frame. Once the parity bytes are calculated, the frame is scrambled to achieve an even spread of 1s, and is placed onto the fiber. At the receiving end, the frame is descrambled and the process reversed for any data terminating at the switch.
Bigger, Better, Faster
So what happens if you want to transmit data at rates higher than DS-3? The answer is the STS-3 frame. It comes in two variations: STS-3 and STS-3c (see "An STS-3 Frame," at above). The STS-3 is simply three STS-1 frames, byte-interleaved. Sonet equipment recognizes an STS-3 by seeing three sets of framing bytes together. These bytes are normally the first two bytes of each STS-1 (see the A1 and A2 in "Overhead Byte Breakdown," below). For all three sets to be together, though, the three STS-1s must be perfectly synchronized. If they weren't, the framing bytes of all three would be scattered, and the Sonet switch would be unable to identify the type of frame. For the three STS-1 frames to be synchronized, the second and third must be offset to match the first. Only the TOH sections are adjusted, and the offset is placed in a pointer to properly identify where the associated SPE is, since the SPE is not adjusted with the corresponding TOH. The three floating SPEs (containing their respective POH and payload) are byte-interleaved and placed into the SPE of the STS-3.
The STS-3c frame was designed to handle even higher data rates than DS-3, such as those from an ATM or FDDI network. The basic difference between the STS-3 and STS-3c frames is the lack of two extra POH columns in the STS-3c. Because the STS-3c frame is composed of data flowing at a higher rate than STS-1, it generally fills the entire SPE and therefore needs only one POH column. There are extra bytes in the TOH also because the STS-3c is not a collection of smaller STS frames. Higher data rates are achieved by byte-interleaving the STS-3 or STS-3c frames.
Because all the frames are interleaved to create larger frames, it's very easy for ADMs to pull individual STS-3s all the way down to individual VTs out of the frame anywhere in the network. With Sonet, there is no need to disassemble the entire frame at each switch in order to pull out, for example, only one DS-1. Because of the byte-interleaving, the switch would know exactly where a particular DS-1 resides. The switch would pull just those bytes out and possibly insert another incoming DS-1 in its place, to be carried down the line to the next switch.
This all combines to create a very fast, efficient, scalable method of data transportation. Sonet was originally created to enable easy data transfers from one vendor to another, but has become a de facto standard for high-speed transmissions. Its design will carry it to even higher speeds and capacities; its only limitation is the number of different wavelengths that can be discerned.
Send your comments on this article to Darrin Woods at dwoods@nwc.com.
|
REPORTS
ANALYTICS
Follow 
