Call Processing Redundancy Issues

One of the most important call processing design attributes for a large number of customers is redundancy. The term redundancy can be an ambiguous term if not properly defined. For example, a PBX with dispersed local controllers can be characterized as a redundant call processing design, simply because failure of one local controller usually has no effect on the other local controllers. A localized failure that does not affect systemwide operations can be considered a form of redundant design because the loss of 100 ports due to local controller failure is not as catastrophic as 100 percent system loss should the Main System Processor fail. Loosely defined, all dispersed control call processing designs are redundant call processing designs. This definition, however, may not satisfy the reliability and survivability requirements of many PBX customers.

If call processing redundancy is more clearly defined by a customer as having a readily available back-up processor element should an active processor element fail, then redundancy requires a duplicate local controller for each local processor element in the dispersed control design example. Duplication of processing elements is a high degree of redundancy.

Regardless of the call processing design category, a PBX system with a fully duplicated call processing design may include any or all of the following:

1. Fully duplicated common control complex, including the Main System Processor, the Main System Memory (software program, customer database), and the Mass Storage Device

2. Fully duplicated local controllers at the port cabinet and/or carrier shelf level

3. Fully duplicated processor bus, including intercabinet communications links

Although the reliability level of the typical PBX system common control complex is very high, usually 99.999 percent (about 5 minutes average annual downtime), hardware and software failures and problems can occur. If there is a problem with the Main System Processor, then all system operations and all system ports can be affected. For this reason, a duplicated Main System Processor is usually required by customers who wish to avoid even minimal service disruptions. In addition to the Main System Processor, customers may request duplicated Main System Memory elements, especially the generic program. A duplicate Mass Storage Device may also be requested because loss of customer database records will affect call processing operations to the same extant as Main System Processor failure.

In a PBX system with a duplicated common control complex, if the active Main System Processor or Main Memory experiences problems, then the back-up (passive) call processing element should instantaneously take control of system operations without interruption of service; active calls remain connected and all activated features continue operating. This is commonly known as a hot-standby duplicated common control system. The only call processing event that is disrupted when the passive element assumes control is a call in the process of being set-up, before call connection to the called party; otherwise, all system functions and operations continue as if nothing happened. If the passive Main System Processor assumes call processing control but all existing switch connections are lost, then it is said to be a cold-standby duplicated control system. A cold-standby system also may require a few seconds or minutes before it is available to begin new call processing operations.

The passive common control elements in a hot-standby design are said to be shadowing the activities of the active elements. For example, the active and passive Main System Processors monitor port status and switch connections, but only the active Main System Processor issues control commands for call processing operations. The passive Main System Processor is merely an observer. Downloads to the active main customer database are simultaneously downloaded to the passive database. Some duplicated common control system designs support operations between the back-up passive Main System Processor and the active Main System Memory and between the active Main System Processor and the back-up passive Main System Memory. This form of shadowing is known as crossover arbitration. Common control complexes with basic shadowing capability transfer all call processing and system operations to the passive processor and memory elements when any of the active common control elements fails; the more advanced shadowing design allows system operations between active processor or memory elements that do not experience problems and the back-up passive processor or memory element—active processor and passive memory or passive processor and active memory. The crossover operation allows for four modes of full function system operation:

1. Active elements (only one)

2. Passive elements (only one)

3. Mix of active and passive (two)

A duplicate common control complex is usually available only in intermediate/large PBX system models. Almost all intermediate/large PBX system models offer duplicated common control as a standard or optional capability. Small PBX system models have traditionally been designed without a duplicate common control complex, even as an option, because manufacturers originally decided that the added system cost to the customer would result in limited sales potential. Likewise, limited sales would not justify the research and development dollars expended for the design. Small system customers who require a fully duplicated common control complex are forced to buy a larger system model to satisfy that need. These customers pay a price penalty for installing a PBX system model with a greater than needed port capacity, because duplicate common control is not available in small system models better suited (and less costly) for their port capacity requirements. For example, the Avaya, Nortel Networks, Siemens, and NEC small system PBX models targeted primarily at customers with fewer than 200 stations are not available with duplicate common control as a standard or an option. Customers must step up to the larger models, sometimes two models above the entry model, if duplicate common control is a requirement. Avaya, Siemens, and NEC offer duplicate common control only as an option on their intermediate/large system models. The duplicate common control option may add as much as 25 percent to the basic system price for small system customers, in addition to the higher cost for the larger system model. Of the four manufacturers, only Nortel Networks offers duplicate common control standard on its intermediate/large system models (Meridian 1 Options 61C and 81C). Figure 1 shows the core module complex of the large Meridian 1 systems.

Figure 1: Meridian 1 option 61/81C core complex.

Dispersed control designs with duplicated local controllers are available in many intermediate/large system models. Most, but not all, PBX systems using a dispersed control design with a duplicated common control complex capability also offer duplicated local controllers. For example, the Siemens Hicom 300H Model 80 can be equipped with duplicated common control and duplicated local control function as an option, but the Nortel Meridian 1 Options 61C and 81C with duplicated common control modules are not available with a duplicated Controller Card. It is possible to have duplicated common control but nonduplicated local controllers. PBX systems equipped with duplicated local controllers are usually available with duplicated common control.

The duplication of the processor bus and intercabinet links is another important redundant call processing system design capability. Processor bus problems can affect call processing operations the same way as Main System Processor or local controller problems. Duplicated processor bus design is inherent to particular PBX system models and is usually available with systems offering a duplicate common control complex. Intercabinet links are less likely to be fully duplicated as a standard design capability, and duplication of the links usually depend on installation of a duplicate common control complex and/or duplicated local controllers. Intercabinet links are part of the system design, whether the call processing design is centralized, dispersed, or distributed, because signaling and communications among the processing elements dispersed among multiple common equipment cabinets may depend on the links for a variety of call processing operations. A single link failure to the Main System Processor may affect hundreds and possibly thousands of system ports housed in the isolated port equipment cabinet. Intercabinet links in a PBX system with a distributed control design may not affect call processing operations in the isolated cabinet, but all intercabinet communications will be affected.