Method for load distribution among the central processors of a multiprocessor central control unit of a switching system

Claim 1

A method for distributing processing load among the central processors (BP, CP, IOC) of a multiprocessor central control unit of a switching system, having a plurality of central processors (BP, CP, IOC) which, in case they are free and operable, respectively automatically fetch a job for processing a switching task, such as a job relating to a detail of "calling" or "through connecting", stored at a central job registering means (AR), fetching it from said central job registering means; and having a central main memory (CMY) which stores data about the current connections in various memory areas and to which the central processors (BP, CP, IOC) have access via a central bus system (B:CMY), said processors being connected in parallel to said central bus system (B:CMY), incIuding the steps of; storing in said control processor a protective code corresponding to the address of this central processor (CPx), attaching such protective code to the data of each and every connection to be set up or be maintained or be cleared down, such code being attached thereto in at least a single, characteristic location as soon as and as long as this central processor (CPx) is handling a fetched job appertaining to this connection, immediately erasing the protective code for the appertaining connection it has entered as soon as the central processor (CPx) has handled its job, whereby each (for example CP9) of the central processors (BP, CP, IOC) which fetches a new job in the job registering means (AR) immediately checks the data stored in the main memory (CMY) which relate to the connection belonging to its (CP9) job, checking these upon subsequent reading to see whether a protective code for the appertaining connection allocated to another central processor (for example CPx) is already stored, and whereby the central processor (CP9) checking the protective code does not handle its (CP9) fetched new job when the protective code is allocated to another central processor (CP9) and does not completely read the appertaining data, characterized in that every central processor (BP, CP, IOC) has its own job memory (AS) individually allocated to it, with those further, new jobs which relate to the same connection which the central processor allocated to the job memory (AS) is currently handling being written thereinto; when a second central processor (CP9), in checking a protective code, identifies that this protective code is allocated to another, first central processor (CPx), the said second central processor (CP9) immediately writes its own, new, second job most recently fetched from the job registering means (AR) into the job memory (AS) of the first central processor (CPx), so that this first central processor (CPx) can itself in turn handle this new, second job, after the conclusion of its (CPx) previous, first job; whereby the second central processor (CP9) can immediately fetch another, new, third job in the job registering means (AR) or, respectively, can fetch another, new, third job from its (CP9) own job memory (AS).

Claim 2

A method for distributing processing load among the central processors (BP, CP, IOC) of a multiprocessor central control unit of a switching system, having a plurality of central processors (BP, CP, IOC) which, in case they are free and operable, respectively automatically fetch a job for processing a switching task, such as a job relating to a detail of "calling" or "through connecting", stored at a central job registering means (AR), fetching it from said central job registering means; and having a central main memory (CMY) which stores data about the current connections in various memory areas and to which the central processors (BP, CP, IOC) have access via a central bus system (B:CMY), said processors being connected in parallel to said central bus system (B:CMY), incIuding the steps of; storing in said control processor a protective code corresponding to the address of this central processor (CPx), attaching such protective code to the data of each and every connection to be set up or be maintained or be cleared down, such code being attached thereto in at least a single, characteristic location as soon as and as long as this central processor (CPx) is handling a fetched job appertaining to this connection, immediately erasing the protective code for the appertaining connection it has entered as soon as the central processor (CPx) has handled its job, whereby each (for example CP9) of the central processors (BP, CP, IOC) which fetches a new job in the job registering means (AR) immediately checks the data stored in the main memory (CMY) which relate to the connection belonging to its (CP9) job, checking these upon subsequent reading to see whether a protective code for the appertaining connection allocated to another central processor (for example CPx) is already stored, and whereby the central processor (CP9) checking the protective code does not handle its (CP9) fetched new job when the protective code is allocated to another central processor (CP9) and does not completely read the appertaining data, characterized in that every central processor (BP, CP, IOC) has its own job memory (AS) individually allocated to it, with those further, new jobs which relate to the same connection which the central processor allocated to the job memory (AS) is currently handling being written thereinto; when a second central processor (CP9), in checking a protective code, identifies that this protective code is allocated to another, first central processor (CPx), the said second central processor (CP9) immediately writes its own, new, second job most recently fetched from the job registering means (AR) into the job memory (AS) of the first central processor (CPx), so that this first central processor (CPx) can itself in turn handle this new, second job, after the conclusion of its (CPx) previous, first job; whereby the second central processor (CP9) can immediately fetch another, new, third job in the job registering means (AR) or, respectively, can fetch another, new, third job from its (CP9) own job memory (AS).