LOW-SPEED VS. HIGH-SPEED 557 There are a number of differences between the low-speed SST and high-speed SS7 technologies. The differences are in the lower two layers of the protocol stack (i.e., the MTP-1 and MTP-2 layers). A comparison of the two technologies is made here. The low-speed signaling data link is a full-duplex, digital transmission channel operating at 64 or 56 kbps. The packets carried using this protocol are variable length and carried down a single clear channel link. This means that the link must be dedicated to the SS7 traffic and cannot be used to carry any other data. MTP-1 is the lowest level of this protocol and is responsible for the physical connection to the transport links. MTP-2, the next layer up, is responsible for transferring traffic between two SS7 components. The initial deployment of high-speed SS7 will operate at a speed of 1.544 Mbps. The U.S. market has adopted the asynchronous transfer mode (ATM) over T1 standard for high-speed SS7, the only exception being AT&T which adopted the clear channel method developed by Lucent. As such, only the ATM version of high-speed SS7 is covered here. With the high-speed SS7 system, the MTP-1 and the MTP-2 layers are replaced. The MTP-2 layer is replaced with the signaling ATM adaption layer (SAAL), and the MTP-1 layer is replaced with ATM. The data is packaged into a CPCS-PDU (common part convergence sublayer-protocol data unit) by adding the service-specific connection-oriented protocol-protocol data unit (SSCOP-PDU) trailer and the ATM adaption layer (AAL) CPCS PDU trailer. (Both are shown in the figure.) This adds eight octets of header detail onto the size of the message. The protocol data unit is then split into 48 octet sections so that it will fit into the payload of the ATM cell. PROBLEMS INVOLVED While one high-speed link can theoretically replace up to 24 low-speed links, in practice the replacement ratio is variable. Where low-speed SS7 is a variable-length packet communications protocol, high-speed SS7 is fixed length. This implies that, depending on the traffic profile, the efficiency of the new high-speed links can vary. Each ATM cell in the high-speed link is fixed at 48 octets, so very short message signaling units (MSUs) are less efficient because they do not utilize the entire available payload. As the length of the MSU increases, the efficiency increases until the point that the MSU becomes too large for one packet and the efficiency drops again. Deciding which links should be replaced can be difficult. The replacement ratio can vary depending on the type of traffic that must be transported. If an incorrect decision is made, the new high-speed link (a) may not be sufficient to cope with the data that it is expected to transport or (b) may be vastly underused. This is why proper profiling of traffic is essential when deciding which links should be replaced.

CALCULATING USAGE ON PROPOSED HIGH-SPEED LINKS Calculating the theoretical usage of the new high-speed links from traffic on the old low-speed links requires a number of steps. Data regarding current usage in the links and link sets that need to be replaced must be collected. This information can then be broken down into the number of 48 octet packets. Due to the fact that high-speed SS7 requires a different length of header than low speed to transport the data, as previously explained, allowances must be made for these extra bytes. An example of this calculation using data collected with the Hewlett-Packard acceSS7 Network Investigator application shows the link set monitored has 12 channels arranged in six links, with each link carrying data at 0.40 erlang. This would be close to the point where replacing the link set would be considered. The channels in this example are running at 56 kbps, which means that they can each carry, at most, 25.2 megabytes in an hour. The expected occupancy of the high-speed link from this can be calculated to be 9% of the total bandwidth. To discover the actual high-speed link occupancy, calculate as follows: Each MSU will have 12 bytes added in the last ATM cell due to the SSCOP-PDU trailer and the CPCS-PDU trailer. From this, the results are split into groupings depending on the number of ATM cells that will be required to transport them. The total number of cells required, over our one-hour period, to transport the data is shown below: = (1 x No. of one-cell MSUs) (2 x No. of two-cell MSUs) (3 x No. of three-cell MSUs) = (1 x 2,990,568) (2 x 949,392) (3 x 138,460) = 5,304,732 cells (1) From this the total number of bytes that would be transmitted on the high-speed link can be calculated: = Number of cells x ATM cell size 281,150,796 bytes/hour (2) The calculation of the total amount of data that could, theoretically, be transmitted over the new high-speed link is as follows. (assuming this is a T1 physical link): = No. of channels x channel speed = 2 x (1,544,000 / 8) x 60 x 60 = 1.39 x 109 bytes/hour (3) This gives a link occupancy of: = the number of bytes transmitted in an hour/maximum transmittable bytes in an hour = (2) / (3) = 0.20 or 20% The occupancy is actually higher than the 9% expected. This is due to the cell tax. The 11% difference actually represents 76.428 Mb per channel per hour. By exporting the results gained from running this measurement to a spreadsheet, these calculations can be manipulated automatically. Such a facility exists in the case of HPacceSS7 Network Investigator. POST-DEPLOYMENT PROBLEMS

High-speed links are not always the right solution because of overloading. Even when highspeed links have been deployed, errors still occur. The need for continuous monitoring to guarantee the continued health of the newly installed links is as great as it was for the low-speed links. Real-time monitoring ensures that should there be any outages or faults within the network, the impact on customers is minimal. The faster the faults can be detected and resolved, the better quality of service for customers. And the more customers using the network, the higher the potential revenues.