Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1998-12-10
2002-08-20
Jung, David (Department: 2175)
Data processing: database and file management or data structures
Database design
Data structure types
C709S241000, C709S241000
Reexamination Certificate
active
06438551
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a load control and overload protection device for a real-time communication system.
BACKGROUND OF THE INVENTION
The purpose of the load control function is to maintain a high flow of successfully handled calls during overflow conditions in a real-time communication system. This is achieved through regulating the accepted call intensity and the number of existing calls that can be disconnected by the real-time communication system taking into account the load situation of a central processor. The load regulation function does not affect the traffic process in normal traffic situations, however, it is intended to prevent an extremely high offered call intensity and/or call disconnection intensity from causing failure of the real-time communication system.
Usually, a call may be subdivided in to several jobs to be handled by the real-time communication system.
As job requests handled by the system may have real-time requirements such as dial tone delay and through connection delay according to ITU-T, the system must be able to handle an overload situation. Overload situations may arise when more jobs are offered per time unit than the system can handle. This means that the system may be heavily loaded and may not be able to accept all offered jobs. In other words, an overload situation can be seen as a peak of offered jobs, that last too long.
However, although a peak of offered jobs depending on the character of the jobs in question, may risk to overload, e.g., job buffers, it does not necessarily mean that the system is overloaded to the extent that real-time requirements can not be kept. To the contrary, the number of offered jobs may overload the system, so that real-time requirements can not be kept, although the job buffers are not overloaded. From this it follows that overload may occur for either any of the job buffers or a central processor CP or both at the same time.
An example for a load control and overload protection system to avoid an overload situation is shown in
FIGS. 32 and 33
and described in Ericsson Review Number 3, 1995. This example relates to the load control and overload protection of a central processor CP handling call related events and other data communication processes in a SPC telephone system. Such a system is designed to handle a prescribed workload of call connections and related jobs.
One problem encountered in this context is how to protect the, e.g., the central processor PC from performance degradation or system failure in case traffic is more intensive than under normal conditions. Examples for such situations are televoting when a large number of subscribers starts making calls simultaneously that are transferred by regional processors RP to the central processor CP or networking failures during busy hour. Here, if no actions are taken, this will eventually lead to congestions and overflow of job buffers JB in the system where jobs are queued and the consequence would be a restart of the system and eventually stop scanning of external requests.
As shown in
FIG. 32
, there is provided a regional processor handler RPH which handles the signalling to and from the regional processors RP. Before the regional processor handler RPH distributes any external signals, i.e. signals from the regional processors, to the corresponding a job buffer in the central processor CP the occupation level of the buffer is checked. This is order not to overflow the job buffers by distributing signals to the job buffers faster than the central processor can fetch signals from the job buffers and execute the job initiated by each signal. If this would be the case the regional processor handler RPH will halt the distribution of signals to the job buffers JB in the central processor CP until the overload situation has ceased.
Equally the regional processor handler RPH will halt the distribution of external signals to the central processor CP for a certain time if the central processor CP has been permanently overloaded for a gross duration of time. This in order to allow the central processor CP to fetch and execute jobs queued in job buffers and eventually cease the processor overload situation.
Thus, the regional processor handler RPH carries out an autonomous flow control of external signals entering the central processor CP with regard to job buffer occupation, i.e. job buffer load. To some extent the regional processor handler RPH also helps avoiding permanent processor overload of gross duration, as an overload protection (not load control).
Signal(s) queued in any of the job buffers will be fetched by the operating system in the central processor CP and distributed to the appropriate application software for execution. As many different optional applications may be running the load control must be made in conjunction with the application software, as only the application knows what type of signal is received, e.g., if it is related to a call already in progress, to the connection of a call or to a new call for which processor load must be requested. For, e.g., messages in a service control point SCP in an intelligent network IN, processor load may be requested for initial, continue and end messages.
Each application will request processor load capacity from a load controlling function. The load controlling function employs a hierarchy of, e.g., 16 call buffers CB where requests may be temporarily stored, if not submitted capacity immediately, until the call set-ups continue. By using several buffers, it is possible to handle calls priorities so that call requests with different priorities are stored in different buffers.
As shown in
FIG. 32
, in case a call request arrives at the regional processor handler RPH it is analyzed to determine which job buffer JB it should be mapped to. In case of buffer capacity the delivery of signals is halted and the scanning of regional processors RP is stopped.
Otherwise, delivered signals and job requests are submitted to the central processor. To ensure that the work load offered to the central processor CP in an overload situation is close to its loadability limit, the number of collected job requests is regulated. Here, the regulation imposes an upper limit on the number of requests collected each time. This limit is adjusted once during every regulation interval, e.g., once every second. In case the load of the central processor CP is above a certain system dependent level the maximum number of collected requests is reduced. Further, in case the load is below a certain system dependent level the limit is again increased. Overall, the object is to keep the load of the central processor cP close to the loadability level during an overload situation.
Thus, load control incorporates mechanisms to keep throughput of requested jobs with real-time demands high under constant or slowly varying overload. To this end, the rate at which new calls are given service is regulated in case the load of the central processor CP is high. Here, in particular jobs corresponding to incoming calls are rejected if the load is too high. Overall, the load control mechanism is designed to be robust in situations involving various types of overload and traffic mixes so as to maintain adequate throughput and short delays in compliance with the telecommunication standard sector of the International Telecommunication Union, ITU-T.
FIG. 33
shows typical load control characteristics. The upper diagram in
FIG. 33
gives a comparison of job throughput against offered traffic without load control (dotted line) and with load control (solid line). As shown in
FIG. 33
, the throughput without any load control will rapidly approach a level of zero in case no precautions are taken, while otherwise a throughput close to the maximum may be maintained. Similarly, as shown in the lower diagram of
FIG. 33
, without load control the delay time will rise considerably while being essentially constant in the case of load control.
One important principle being related to load control is called back pressure o
Burns Doane Swecker & Mathis L.L.P.
Jung David
Telefonaktiebolaget L M Ericsson (publ)
LandOfFree
Load control and overload protection for a real-time... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Load control and overload protection for a real-time..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Load control and overload protection for a real-time... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2948342