Previous Table of Contents Next


The virtual path identifier of a virtual path entering the ATM switch is different from the virtual path identifier of the virtual path exiting the switch. Likewise, the virtual channel identifier of a virtual channel entering the ATM switch is not the same as the virtual channel identifier of the virtual channel exiting the ATM switch. Figure 6-11 shows two VPs, called VPI 1 and VPI 5, each composed of three VCIs (VCI 1, VCI 2, VCI 3), entering an ATM switch. VPI 5 is passed through the switch unchanged. VPI 1 is unbundled in the switch and the individual VCIs are switched to other VPI bundles. Figure 6-11 is a simplified representation of the inner working of the ATM switch. VPI 2, VPI 3, and VPI 4 include other switched VCIs from other VPIs in addition to the ones pictured. The ATM switch software includes routing tables that interpret the VPI/VCIs and switch the cells to the correct output, ensuring the cells go on to the correct destination.


Figure 6-12.  Three-port bidirectional ATM switch

The VP path portion of the ATM switch will switch whole VPs from switch ingress to switch egress. The VC path portion of the ATM switch will break down the VP bundles and switch the individual VCs.

The VPI/VCI value has only "local significance," meaning it is useful only to the switch that the value is currently associated with. Each switch in the transmit path changes the VPI/VCI according to its own needs. Figure 6-12 depicts a three-port, bidirectional ATM switch with the ports identified as Port 1, 2, and 3. Port 1 has two cells with VPI/VCI set to 15 and 32. Port 2 has a cell identified with VPI/VCI of 27. Port 3 has a cell identified as VPI/VCI 48. Since the switch depicted is bidirectional, cells can enter and leave each port in either direction. So, cells entering Port 1 are assigned VPI/VCI values of 15 and 32. Likewise, cells leaving Port 1 are assigned VPI/VCI values of 15 and 32. Cells entering or leaving Port 2 are assigned a VPI/VCI value of 27 and cells entering and exiting Port 3 are assigned a VPI/VCI of 48. The table included with Figure 6-12 shows all the possible ingress and egress combinations the three-port switch can manage. It is important to understand the VPI/VCIs assigned to cells coming into and exiting the switch are unique to this switch. The next switch in the transmission path will assign its own VPI/VCIs necessary for it to manage the cells.

There are several VPI/VCI combinations that are of special interest. Signaling cells are indicated when the VPI equals 0 and the VCI equals any positive integer (positive whole number). Empty cells are indicated when both VPI and VCI equal zero.

Header Error Check

The ATM Header Error Check (HEC) code is derived from the polynomial 1 + x + x^2 + x^8. In the HEC polynomial, x refers to the number of ones bits in the cell. The source counts the ones bits, calculates the value of the polynomial, and places the value in the header. The destination calculates a new HEC based on the number of ones bits received. If the value transmitted agrees with the calculated value by the destination, then all is well. If the values do not agree, the algorithm used allows the destination to correct single bit errors or to detect multiple bit errors.

Provisioning and Signaling

An ATM circuit can be either connection oriented or connectionless. A connection-oriented circuit is one that is established between two or more network entities using signaling between the entities. A connectionless-oriented circuit is a circuit that is "nailed up" between two or more network entities. "Nailed up" is telecom lingo meaning the entities of concern are always connected to each other and no switching or signaling is required for one entity to communicate with another.

Provisioning in ATM parlance means a circuit is "nailed up." That is, a provisioned circuit is a circuit that is hard wired along a transmission path from source to destination. There is no switching involved to get from source to destination.

Signaling is synonymous with switching in ATM. The switching function is performed by the ATM switches using the signaling information provided by the ATM cell header.

Sustainable Cell Rate and Peak Cell Rate

Sustainable cell rate (SCR) is a transmission rate defined as the average cell rate that may be transmitted over the backbone. The SCR is measured in megabits per second. The operative word in this definition is "average." The average is taken over some time period. The average time period depends upon the ATM switches deployed by the carrier in their network and the switches’ specific capabilities. Typically, the average time period is taken over either 30-minute periods or 15-minute periods. Cells exceeding the SCR are tagged Discard Enable (DE).

Peak cell rate (PCR) is a transmission rate defined as the maximum allowable cell transmission rate. PCR is measured in megabits per second. In a perfect world, all cells transmitted above PCR are discarded. Guaranteed. However, we do not live in a perfect world and carriers are attempting to attract ATM business to fatten the lightly loaded ATM backbones. So, most carriers are allowing users to burst above PCR, up to the access speed, without penalty, as long as network congestion remains under control. Now is the time for all good men and women to get maximum bang for the buck.

Technically, SCR and PCR are applicable to VBR services, which are bursty by nature. Most carriers, however, are going to impose an SCR and PCR of some value on all classes of service to manage system capacity. The only sane way the carrier can manage the carrier backbone is to have a known SCR and PCR for every user. That way, the carrier’s capacity planning departments can, hopefully, actually plan the capacity of the backbone network. Granted, the PCR may be the access speed of the connection but whatever it is, it must be defined for every user, else havoc will reign, eventually.

The challenge for the user is to select the proper service class for the user’s specific network requirements. Selecting the proper service class can still result in poor system performance and/or wasted resources if the correct SCR and PCR are not chosen. The network manager who knows his network data transmission rates can make wise choices.


Previous Table of Contents Next