WO2009122071A2

WO2009122071A2 - Method for managing device load in a system for implementing an application service

Info

Publication number: WO2009122071A2
Application number: PCT/FR2009/050419
Authority: WO
Inventors: Mohamed Boucadair; Yoann Noisette
Original assignee: France Telecom
Priority date: 2008-03-17
Filing date: 2009-03-13
Publication date: 2009-10-08
Also published as: FR2928801A1; WO2009122071A3

Abstract

The invention relates to a load management method for a system for implementing an application service in an IP telecommunications network, said system including a plurality of devices, including border devices for exchanging messages for establishing application sessions with service access networks, characterized in that said method includes the following steps: obtaining an effective load rate of at least one device of the system; if the effective load rate of said at least one device reaches an authorized load rate of said at least one device, calculating a new authorized load rate for at least one of the border devices, said effective load rate of said at least one device being a function of said authorized load rate for said at least one border device.

Description

Method for load management of a device of an implementation system of an application service

Field of the Invention The field of the invention is that of telecommunications networks, and more particularly that of a system for implementing an application service in a telecommunications network.

The invention finds its application for the management of the equipment load of a system implementing an application service in a telecommunications network, for example a telephony service deployed over an IP network. the management of the phenomena of the overload more particularly the problem known as "Flash Crowds" Previous Art The IP network (Internet Protocol) becomes more and more the universal support of a multitude of services and applications It is the network backbone adopted by the Operators to mutuahse their heterogeneous service offerings Among these services, IP telephony or voice over IP, also known as VoIP for Voice over IP, has in recent years taken a prominent place, to the point that all providers of services now have IP offerings for their customers, whether they are residential customers or businesses These service offerings, based on the IP protocol, are opposed by their design and their philosophy to the traditional telephony which relies on the commutation of the circuits the advantage advance for the passage with the telephony on IP, besides the lower cost of the communications, is the capacity estimated to be able to link services with the world of computing and the Internet in particular

To accompany the deployment of VoIP services, I ¹ IETF (Internet Engineering Task Force) carried many standardization work Several protocols were thus specified among which we can mention SIP (Session Initiation Protocol) SDP (Session Description Protocol), RTP (Real-time Transport Protocol), Real-time Transport Control Protocol (RTCP), Multimedia Gateway Control (MGCP) Protocol), SAP (Session Announcement Protocol) and TRIP (Telephony Routing Over IP), etc. These protocols respond to different needs and include, in particular, signaling and call control, the exchange of media streams, their control and the exchange of call routing information. To satisfy the customers they have attracted, telecom operators providing VoIP services must meet various constraints related in particular to the availability of the service, its configuration (provisioning), its cost, the security of the communications flows, fault tolerance and service flexibility. In particular, they must guard against the phenomena of overload that may affect the equipment involved in the implementation of the IP telephony service.

By overload is meant the inability of the service or equipment implementing the service to process a request. This inability is often due to a lack of available resources when receiving the request. The management of an overload is all the more delicate, it is a phenomenon that tends to maintain itself.

Indeed, the problem often affects progressively the other equipment of the service, which trying to sell the traffic, and are in turn overloaded. In addition, a portion of the resources is then used to send error messages or alerts, which further impinges on the resources available to solve the problem. An additional degrading factor is the retransmission of requests that is intended to prevent potential IP transport errors. Such a reissue procedure is described, for example, in the RFC defining SIP [RFC3261] Disadvantages of the Prior Art

Engineering rules allow telecom operators to avoid this kind of phenomenon under normal operating conditions, on the one hand by correctly dimensioning their service, on the other hand by implementing load-sharing mechanisms (or load balancing in English) between the nodes constituting the service. However, these mechanisms are for the most part insufficient time for exceptional events that lead to peak loads (New Year examples, earthquakes, power cuts, serious events such as those that occurred during the events of 11 September, etc.) or even a failure affecting the customer base or the service platform

One of the less obvious overload phenomena to be understood is the so-called "Flash Crowds", which typically occurs when a large number of users simultaneously attempt to make a call, as the service elements have not been sized to absorb this influx simultaneously (ie traffic burst) Indeed, the sizing of the telephony networks is notably based on a statistical distribution of the calls, governed in particular by the laws of Erlang Or this type of dimensioning is not destined to result in the deployment of a service that can cope with occasional traffic overloads In the case of a "Flash Crowds" type phenomenon, the origin of a bottleneck at the level of the service platform can then emanate from two particular situations (1) a large influx of calls to a particular destination (eg phone number or geographical area) or (2) a large influx from a region n (in terms of connection to the service) given An example is to consider a game show announcing a phone number to join to receive a sum of money This kind of situation can lead to a large number of calls to the number in question. question over a relatively short period, resulting in congestion degrading the level of service rendered and therefore that perceived by users (other than those calling the said number)

A known solution of the prior art and described in the document RFC 3261 of the SIP (Session Initiation Protocol) is that the congested equipment notifies the client of the impossibility to access the service. purpose the error code in its response "503", sending a SIP error message with this code meaning that the requesting service element is not able to process the request following a temporary overload or a maintenance operation The equipment in question can indicate on this occasion a delay after which the customer can make a new attempt (especially with the field "Retry-After"). A customer receiving such a response code is expected to request other equipment to fulfill his request, at least until the time specified in the "Retry-After" field (if indicated) and if the opportunity to contact a other equipment is offered

An alternative proposal is still under discussion within the standardization bodies, in particular the Internet Engineering Task Force (IETF). It consists of the addition by the equipment of the service, in the signaling messages, for example according to the SIP ("Session Internet Protocol") protocol, of a field dedicated to the treatment of the phenomena of congestion / overload, in order to provide a better qualification of the state of congestion. This dedicated field of the SIP header, called "CONGESTION", can be inserted into any SIP response. It allows in particular to characterize the degree of congestion found (value ranging from 0 to 4), the type of congestion (eg CPU, network, etc.), the equipment concerned as well as a delay beyond which a new attempt can be made. be carried out (assuming the degree of congestion is at its maximum). This solution is similar to using the "Retry-After" tag described above because the service equipment must generate a message and execute a procedure to determine (see predicting, given present and future demands) the deadline its return to normal, which is not easily achievable during exceptional overload phenomena (such as crises, important events, etc.).

A first disadvantage of these solutions is that they assume that the congested equipment is able to generate such a message and proceed with the associated processing when overloading the service. A second disadvantage is that they generate additional traffic. A third disadvantage is that they solve the overload problem curatively. Concretely and in addition to the disadvantages listed above, we can also mention the following problems:

Amplification of service overload Refocusing a customer to another service equipment, which itself may be overloaded, may induce a significant increase in number of messages exchanged, increasing the overall overhead can even lead, in the extreme, to a collapse (ie unavailability) of the service This technique is not adapted from the moment when several equipment are in overload situation

Underutilization of resources The scope of a SIP message with error code "503" is not sufficiently precise Typically, if several servers are grouped under the same domain name, and one of them is overloaded, a client could consider that it is not necessary to solicit the domain name in question, whereas in reality only one server among those pointed by the domain name is overload With the heading "CONGESTION", one minimizes somewhat this problem, because this header contains a field allowing to characterize more precisely the element having sent the message However, the exploitation of this information is not described in the documents of specification, and can thus be different from one implementation to another

Lack of flexibility The notification of a delay thanks to the field "Retry-After" allows an overloaded equipment to indicate to the requester to emit the request only at the expiration of a "timer" gives Moreover, overloaded equipment may or may not redirect all the traffic that arrives to it, with no possibility of redirecting only a portion after the occurrence of the overload phenomenon Usage Ambiguities The procedure described in RFC3261 does not sufficiently specify the circumstances under which the response error code "503" must be used As a result, other use cases have been specified or even implemented. This complicates the precise interpretation of this code for its own purposes of overload detection.

Presentation of the invention There is therefore a need to overcome these drawbacks of the prior art. More specifically, there is a need to provide a solution for preventing an overload phenomenon in a system for implementing an application service hosted by a telecommunications network. invention responds to this need by using a load management method of a system for implementing an application service in a telecommunications network, said system comprising a plurality of equipment including border capable of relaying application session establishment messages between equipment of said system and service access networks

According to the invention, the charge management method is characterized in that it comprises the following steps

obtaining an effective load rate of at least one equipment of the system,

- if the effective load rate of said at least one equipment reaches an authorized load rate of said at least one equipment, calculating a new load factor allowed for at least one of the edge equipment, the effective load rate of said at least one an equipment being a function of said authorized load rate for said at least one edge equipment

Thus, the invention is based on a completely new and inventive approach to load management supported by a system for implementing an application service in a telecommunications network. Indeed, the invention proposes a management mechanism. the burden of correlating the load ratio of at least one system equipment to the individual load rates of the curbside equipment of this system. A control of the individual load rates of the curb equipment allows to control the loading rate of at least one equipment of the system, so that its effective load rate remains below an authorized load rate

A fundamental criterion adopted by this invention is no longer the detection of the overload of a service equipment, but the prevention of an overload of the system, by relying on the actual load rates measured at the periphery or the core. of system The implementation of the invention is done by imposing load thresholds, or allowable load rates, not to be exceeded, at the edge equipments which allow the interconnection with third party operators or with networks of access to the network. The invention thus makes it possible to solve the technical problem of preventing overload phenomena such as the "Flash Crowds" phenomenon.

According to one aspect of the invention, the authorized load rate for a border equipment is calculated from an acceptance rate of received signaling requests. It is indeed considered that the media data traffic generates can be deduced from the messages. signaling, for example from a priori knowledge on the consumption of resources of a particular CODEC

An advantage of this aspect of the invention is that it dynamically adjusts the incoming traffic at the edge equipment according to the state of charge of the system in general and the edge equipment in particular. first embodiment of the invention, the effective load ratio obtained relates to a plurality of equipment forming a centralized platform

This rate constitutes an overall effective load ratio of the centralized platform, which is considered, according to the invention, as a function of the individual effective load rates of the system edge equipments. An advantage is to control this overall load rate by regulating the individual load rates of the edge equipment

According to a second embodiment of the invention, the effective load rate obtained relates to a border equipment and the step of calculating an authorized rate evaluates the possibility of assigning to said edge equipment a new authorized rate higher than the authorized current and less than maximum capacity of said border equipment

One advantage is that, when a large influx of signaling requests at a level of edge equipment, for example when a crisis situation occurs in a given region, the method according to the invention gives the possibility to this equipment. to absorb the overload by increasing it to the extent of its capacity, the value of its maximum permitted load

Advantageously, the method further comprises a step of obtaining an effective load rate of a plurality of equipment forming a centralized platform and in that the calculation step is implemented if the load ratio the centralized platform is less than an authorized load rate for that platform

One advantage is that the allowable load rate for the edge equipment is increased only to the extent that the permitted overall load rate of the service platform allows it. According to yet another aspect of the invention, the method of charge management comprises a step of sending the new rate authorizes said at least one relevant border equipment

The border equipment concerned can therefore take into account this new rate for the processing of the signaling requests it receives. According to another aspect of the invention, the step of obtaining an effective load rate of the platform centralized method comprises a step of recovering said charge rate from a supervisory equipment

According to this aspect of the invention, the overall charge rate of the telephony system is not estimated by the equipment that implements the method, but recovered from one or more other equipment. One advantage is not to overload the telephony system equipment that implements the load management method and to take advantage of load measurements that are generally available in the heart of a telephony system.

Alternatively, the charge management method according to the invention comprises a step of estimating the overall effective load ratio of the telephony system from said at least one piece of information obtained (from the edge equipments).

According to this aspect of the invention, the equipment which implements the charge management method according to the invention is capable of estimating the overall system load from the individual loads of the edge equipments. the invention is based on the fact that the effective loading rate of the service platform can be mathematically related to the individual effective load rates of the edge equipment. An advantage of obtaining the effective load rate of the service platform by calculation, is not to depend on a supervision equipment One advantage is that a border equipment can implement the method according to the invention without depending on another equipment of the heart of the system

According to another aspect of the invention, the charge management method further comprises a step of obtaining an authorized charge rate of said edge equipment. The current authorized charging rate of the edge equipment is therefore transmitted by the edge equipment itself to the equipment that implements the invention at the same time as the effective load rate This aspect of the invention is particularly advantageous when the load management method is likely to be put implemented by at least two devices of the application system, so that one can be aware of a load rate authorized by the other This is for example the case, when the method according to the invention is implemented by A further advantage is the prevention of a failure of the configuration equipment by ensuring that other equipment in the system, which can replace it if needed, has the same information. In a particular embodiment, the various steps of the load management method of an implementation system of an application service are determined by computer program instructions.

Consequently, the invention also relates to a computer program on an information carrier, this program being capable of being implemented in a configuration equipment of the telephony system or more generally in a computer, this program comprising instructions adapted to the implementation of the steps of a configuration process as described above

This program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other desirable form

The invention is also directed to a computer-readable information medium, comprising instructions of a computer program as mentioned above. The information medium can be any entity or device capable of storing the information. For example, the medium may comprise storage means, such as a ROM, for example a CD-ROM or a microelectronic circuit ROM, or a magnetic recording medium, for example a floppy disk or a diskette. On the other hand, the information carrier may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The program according to the invention can be downloaded in particular on an Internet-type network

Alternatively, the information medium may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the method in question

The configuration method according to the invention can in particular be implemented by a configuration equipment of the implementation system of the application service. Therefore, the invention relates to a configuration equipment of a system for implementing an application system. application service in a telecommunications network, said system comprising a plurality of devices including edge equipment able to exchange application session establishment messages with service access networks, characterized in that said configuration equipment includes the following means

obtaining an effective load rate of at least one equipment of the system,

- if the effective load rate of said at least one equipment reaches an authorized load rate for this equipment, calculating a new load factor allowed for at least one of the edge equipment, the effective load rate of said at least one equipment being a function of said authorized load rate of said at least one equipment border.

According to this centralized embodiment of the invention, the method is implemented by a configuration equipment, to which a plurality of edge devices return their charge rate information. The configuration equipment calculates the new allowed rates for the system edge devices and sends them for enforcement. An advantage of such a centralized embodiment is the control of the system behavior and the ability for the service operator to modify the service configuration by acting on the central element implementing the load management method. Another advantage of this solution is to guarantee the consistency of the service rendered.

The invention also relates to a method for regulating the load of a border equipment of an implementation system of an application service in an IP telecommunications network, said system comprising a plurality of equipment including edge capable of exchanging requests for establishment of application sessions with service access networks, characterized in that it comprises the following steps: sending an effective load rate to a configuration equipment of said system; receiving a new authorized load rate from said configuration equipment; and - regulating a processing of signaling requests received by said edge equipment according to the new authorized charging rate

According to one aspect of the invention, the sending step is implemented when said effective load rate reaches the authorized load rate. One advantage is to solicit the configuration equipment only when an event justifies it. According to one variant, the sending step is implemented according to a predetermined sending rhythm.

One advantage is to regularly adjust the authorized load rate of edge equipment to traffic. In a particular embodiment, the various steps of the charge control method are determined by computer program instructions.

Consequently, the invention also relates to a computer program on an information medium, this program being capable of being implemented in a system edge equipment or more generally in a computer, this program comprising adapted instructions implementing the steps of a configuration method as described above.

The invention also relates to a border equipment of a system for implementing an application service on a telecommunications network, said system comprising a plurality of equipment including edge equipment capable of relaying establishment messages. application sessions between system equipment and service access networks, characterized in that it comprises the following means: sending an effective load rate to a configuration equipment of said system;

receiving a new authorized load rate from said configuration equipment, and

- Regulating a processing of signaling requests received by said edge equipment according to the new authorized load rate. The edge equipment may be, for example, a Session Border Controller (SBC) or a conventional proximity server ( still called Proxy Server according to the SIP protocol).

According to another aspect of the invention, the edge equipment further acts as a configuration equipment according to the invention. According to this distributed embodiment, the edge equipment receives the actual load rates from other edge equipment and calculates new load rates allowed for such equipment, such that the overall load rate of the service platform remains below an authorized global charge rate. One advantage is not to overload the equipment of the core of the system Finally, the invention relates to a system for implementing an application service in an IP telecommunications network, said system comprising a plurality of devices including edge equipment capable of relaying application session establishment messages between system equipment and service access networks; characterized in that it comprises: at least one configuration equipment according to the invention and in that the edge equipment comprises the following means: sending an effective load rate to a configuration equipment of said system; receiving a new authorized load rate from said configuration equipment; and - regulating a processing of signaling requests received by said edge equipment according to the new authorized charging rate,

List of Figures

Other advantages and characteristics of the invention will emerge more clearly on reading the following description of a particular embodiment of the invention, given as a simple illustrative and nonlimiting example, and the appended drawings, among which: FIG. 1 schematically shows a system for implementing an application service in a telecommunications network according to the prior art; FIG. 2 presents the steps of the load management method of a system according to FIG. invention; Figure 3 shows the steps of the load management method of a system according to a centralized embodiment of the invention; FIG. 4 presents the steps of the load management method of a system according to a distributed and hierarchical embodiment of the invention; FIG. 5 presents the steps of the load management method of a system according to a distributed and non-hierarchical embodiment of the invention, FIG. 6 schematically shows the structure of a configuration equipment implementing the method. load management system according to the invention; FIG. 7 schematically shows a system for implementing an application service in a telecommunications network according to a centralized embodiment of the invention, FIG. 8 schematically shows a system for implementing a application service in a telecommunications network according to a distributed and hierarchical embodiment of the invention, and FIG. 9 schematically shows a system for implementing an application service in a telecommunications network according to a centralized embodiment. and non-hierarchical of the invention

Description of a particular embodiment of the invention

The general principle of the invention is based on the control of the overall load rate of a service architecture deployed on a telecommunications network, a system for implementing an application service, by regulating the load of the equipment constituting this service. system, in order to avoid reaching a critical load threshold Such regulation is done at the level of the edge equipment located in the telecommunications network hosting the system Indeed, it is these edge equipment which, by allowing the exchange signaling and media flow requests between the application service implementation system and an access network through which a user terminal accesses the service, sees the incoming traffic in the system

The function of the edge equipment can be rendered by dedicated equipment or supported by a proximity server (or called Proxy Server) when the implementation of the service is done through the collaboration of a plurality of telephony platforms. a plurality of edge equipments is adopted in particular in most of the IP telephony services architectures deployed to date, according to which the clients' access to the conversational services is carried out through points of presence (Point of presence). Presence or

POPs), most of the time distributed, which themselves access a centralized service platform, also known as the heart of the service In the remainder of the description, it is considered, for example, a telephony system It will be noted, however, that the invention applies to any type of application service architecture, provided that it requires an application session establishment between the client and the system A CN telecommunications network hosting a telephony system is presented with reference to FIG. 1. A simplified view of this system makes it possible to break it down into three segments the end customers represented in this example as T1 terminals. at T4, service access equipment such as EB1 to EB4 (Session Border Controllers or SBC), DHCP servers or DHCP relay clients that may be located in PoPs (Point of Presence Presence) of given accesses and core network service facilities forming a centralized PFT service platform, such as Proximity Servers (or Proxy Servers) or else database servers (not shown) The edge equipments EBi, i integer between 1 and 4, are able to receive signaling and media messages from a client terminal Ti and to transmit it to the system. In general, the end customers T1 to T4 are distributed (or attached) between different POPs (Point Of Presence) or access networks AN1 to AN4 The edge equipments EB1 to EB4 serve as interface for the signaling and media exchanges between the telecommunication core network and the access networks AN1 to AN4 The edge equipments render several other functions such as, for example, a control of the quality of the service, a control of the codes used or even a protocol correction (for example concerning a validation of the messages, a reformatting of the messages or an addition of specific headers)

In the remainder of the description, it is considered that the signaling messages exchanged comply with the SIP protocol (Session Initiation Protocol). However, it should be noted that the scope of the invention is not limited to this particular case, but relates to any other signaling protocol for the establishment of sessions, such as, for example, the IAX protocol (Inter-Asteπsk Exchange) The charge management method of an implementation system of an application service according to the invention will now be described with reference to FIG. 2. Such a method comprises a step E1 for obtaining GET of at least one effective NPi or Ni charge rate of at least one of the equipment of the IP telephony system, Npi being the effective load rate of a central equipment and Ni the effective load factor of an EBi edging equipment It will be noted that such information can be obtained in PULL mode or in PUSH mode as well. In PULL mode, it is obtained on request from the equipment, whereas in PUSH mode, it is spontaneously emitted by this equipment, by configuration or when a triggering event occurs. The actual charge rate obtained may consist of one or more different types of information II may be, for example, a CPU load measurement. If the service equipment in question is an edge equipment EB1 to EB4, the effective load rate Ni is obtained in a step E11 and concerns at least one state of the active simultaneous voice over IP sessions or a number of access requests. received during a given period of time. According to the SIP protocol, such an access request takes the form of an INVITE message. It will be noted that such a measurement of the number of access requests to the service is a particular trigger of the load management mechanism according to the invention. because the phenomenon "Flash Crowds" that one seeks to prevent is precisely related to an avalanche of requests for access to the service. However, the information obtained may advantageously be of various types, among which there will be, non-exhaustively a number of simultaneous sessions active; a rate of service access request processing, a service equipment CPU occupancy rate or any other relevant information for estimating the load of a service equipment

According to one aspect of the invention, the obtaining step is a step EI ₂ GET-N obtaining an overall effective load rate N of several equipment of the system forming the telephony platform PFT at a given instant t It should be noted that, in a known manner, the current telephony platforms have a supervision system capable of estimating the overall load of the platform from measurements. operating instructions provided by SNMP MIB counters or RMON probes

In a step ES ₁ , the effective load rate Ni for the EB edge equipment is compared to an authorized load rate Mi for this edge equipment.

In a step E3 ₂ , the effective overall charge rate N obtained is compared with a maximum overall charge rate allowed. It will be noted that the authorized charge rate values are generally defined by the service operator who specifies the maximum limits of the charge. load supported by the PFT telephony platform as a whole and, in particular, by each of the EBi border equipment

The method according to the invention is based on the following principles. Considering a ratio n beyond which the telephony platform PFT is considered overloaded. The charge rate Mi is the charge rate allowed, determined by configuration, of the EBi II edge equipment represents the maximum load that this edge equipment is allowed to accept in order not to overload the PFT platform By default, considering a statistical distribution of calls over time, governed by the Erlang laws, the Mi values can be so configured that M = g (Mi (l≤i≤m)) or g (Mi (l≤i≤m)) <n * M (1)

wherein g () is a function that relates the maximum allowable total load rate of the system based on the maximum allowed local load rate of each edge equipment EBi located at the periphery of the service.

We now consider the effective load rate Ni generated by an access POP or EBi edge equipment towards the telephony platform PFT The principle of the invention is to ensure that the following inequality is always verified

N = f (Ni (l≤i≤m)) <n * M (2)

where f () is a function which relates the overall system load ratio as a function of the local load rate of each edge equipment EBi located at the periphery of the service. For example, the function f () can be express as follows f () ≈ Σ Ni (l ≤ i ≤ Ξ m), in the case where one and only one edge equipment EBi is involved in the establishment of the call and in the media flow exchanges, f () = 1 / 2 * Σ Ni (l≈ζi≈≤m), in the case where two edge equipments EBi ₁ EBj intervene in the establishment of the call and in the exchanges of media flows f () = 1 / p * Σ Ni (1 ^ i ^ m), in the case where a number p of edge equipments are involved in the establishment of the call and the exchanges of media flows

According to the invention, a step E4 for calculating a new permitted load rate Ni for at least one edge equipment EBi is triggered, at least in the following two cases

1- if the overall effective load rate N exceeds the authorized global load rate M, that is to say if one is in the situation of a large influx of requests at the level of the centralized platform, which generates a state high load of this platform, or

2- if the effective load rate Ni of the edge equipment EBi is greater than its authorized load rate Mi, while the overall effective load rate N exceeds the authorized total load ratio M, then we are in the situation a large influx of requests for a particular access to the platform, which locally generates a high load of EBi edge equipment.

Preferably, the charge management method according to the invention modifies the configuration of an edge equipment EBj, j 1 to 4, such that the new rate allows M 1 to satisfy the preceding inequalities. In the first case, it is preferentially the edge equipment EBj that has received the largest number of access requests to the service that is chosen. However, it will be noted that the choice of the border equipment (s) concerned by the configuration change depends on the policy of the user. operator and perhaps any other one can consider, for example, a change distributed over several border equipment

The modification decided by the method according to the invention is, for example, to lower the maximum authorized load rate M] for this equipment EBj so that the telephony platform is less stressed by this edge equipment and therefore less charged

It is understood that the edge equipment EBj recipient of the new maximum permitted load rate Mj according to the invention is not necessarily the edge equipment EBi from which the information of effective load rate Ni have been obtained.

In the second case, it is preferentially the edge equipment EBi having emitted the information representative of its effective load rate Ni which is chosen as the recipient of the configuration modification II will evaluate the possibility of increasing its maximum authorized load rate Mi, so that it can absorb the load caused by the processing of the requests it receives while satisfying the previous inequalities

It is thus clear that the step E4 of the charge management method according to the invention directly correlates the load of an edge equipment through which a portion of the incoming request stream into the telephony system is transmitted to that generated in the system of the invention. telephony itself

Advantageously, the maximum authorized load threshold is expressed and is calculated at least according to a maximum number of incoming access requests allowed. However, it should be noted that other parameters may be used for its evaluation, such as the rate of treatment. Incoming requests Therefore, lowering the allowable load rate Mi for EBi edge equipment can result, for example, in a decrease in the acceptance rate of incoming requests and / or in the processing rate of the lowest priority requests.

On receipt of the new authorized load factor Mj, the edge equipment EBj implements the load regulation method according to the invention. Such a method comprises a step of regulating a processing of received signaling requests. This step consists in checking that the processing of the current requests does not cause the equipment a load N] higher than the authorized load rate M].

This is, for example, to dynamically control the number of sessions maintained This call processing capacity is controlled either locally (through a configuration interface) or controlled (controlled) by an external entity The control mode depends on the adopted architecture (ie centralized or autonomous)

The implementation of the load management method of an implementation system of an application service according to the invention can be divided into two main modes.

A centralized mode, according to which the charge management process is executed by a central equipment, for example located in the centralized telephony platform PFT, which receives information on the charging rates of the edge equipments and recovers the overall charge rate of the platform, for example, with its supervision system This equipment does not intervene necessarily in the provision of the service in question An advantage of this mode is to allow the operator of the service a centralized control of the service behavior and a single point of entry from a centralized configuration equipment whose responsibility is to control the border equipment and modify their behavior,

A distributed mode (also known as stand-alone mode) in which edge equipment exchanges information about their actual load rates, determines and regulates the influx of traffic they are able to accept, in such a way that all the signaling and media flow exchanges that they send back to the service platform generate at the platform level a load which remains lower than the maximum authorized load rate M They therefore implement the Load management method according to the invention The distributed mode offers a flexibility, "dynamicity" and flexibility in managing the overload phenomenon without introducing new elements and interfaces

FIG. 3 shows the implementation of the load management method according to a centralized embodiment. The steps E'1 to E'4 are identical to the steps E1 to E4 of FIG. not described

In a step E'5, the new authorized load rate M] is transmitted to the edging equipment EBj concerned According to the invention, said equipment EBj updates its rate and activates traffic regulation rules so as to respect the new charge rate M] not to be exceeded

FIG. 4 shows the implementation of the charge management method according to a distributed embodiment. According to this embodiment, at least one edge equipment EBk, k integer between 1 and 4, implements the charge management method according to the invention II receives, in E ¹¹ I ₁ , information relating to the Ni charge rate of at least one edge equipment EBi of the telephony system.

Advantageously, it also receives information relating to at least the authorized load rate Mi for this equipment

The edge equipment EB _k implements a step E "2 ₂ for estimating the effective global load rate N of the platform from the effective load rates of the edge equipments EB1 to EB4. However, it is also possible to use other ST-INF information stored in memory. This estimate is based, as before, on the principle that the overall effective load of the platform PFT is a function of the effective loads of the edge equipments The following steps E "31, E" 32 and E "4 are identical to the steps E'31, E'32 and E'4 of FIG. not described The distributed embodiment is divided into two variants

in a first variant, all the edge equipments EB1 to EB4 implement the method according to the invention. At the end of step E "4, they therefore all have in their tables the same updated values of the allowed load rates M] for each of the equipment EBj, so they do not have a priori need of mutually exchange the calculated values M] However, for reasons of safety and prevention, it may be advantageous for the border equipment to exchange the new authorized load rates M 1.

in a second variant, a hierarchy is set up within the community of EBi edge equipments. Border equipment is elected as master to implement the charge management method according to the invention Such an election can be done administratively by decision of the operator of the platform, or dynamically. A dynamic election may, for example, consist of a border equipment that has sent information informing Ni and / or Mi about it to its peers and that has not received any response back after a predetermined time self-proclaims master of the border equipment community According to one variant, a border device sends, when it is put into service, a specific request (Who_l S_Master ()). This request is received by all edge equipment. If a master has already been elected / appointed, then the master responds to the originator of the request. The transmitter then indicates in a dedicated table the identity of the master. Only decisions received from this border equipment will be considered. A negotiation phase of a security association may be considered between the master border equipment and the emitter edge equipment. If no master border equipment is present in the system, then the emitter self-proclaims master and sends a message to the community notifying them of this election. Other edge devices then update their tables with the identity of the new master border equipment. Any master edge device may forego this function if, for example, it reaches a critical load, if the decision-making procedure degrades its performance. or for scheduled maintenance operations To do this, it sends a message called TEAR_MASTER () to the group multicast G Once received by the other equipments of border, a procedure of replacement of this equipment of border master is then launched This election is done on the basis of information such as the capacity of treatment of a piece of equipment the current load, the element with the longest IP address, etc. Based on this information all the edge equipments update their tables by indicating the identity of the new entity acting as a device. It sends a confirmation message and triggers a negotiation phase of security associations with its peers. The master edge equipment implements the method according to the invention and in a step E "5, it sends to the border equipment concerned with the new authorized load rates M] that it has calculated These implement the new configurations received

The charge management method according to the invention is therefore implemented by at least one equipment of the telecommunications network CN

According to a first embodiment, called centralized mode, this equipment is a central configuration equipment, constitutive or not of the centralized platform PTF, such as a server. According to a second embodiment, called distributed mode, the management method of load is implemented by at least one edge equipment EBi In the case of the hierarchical distributed mode, the calculation and decision-making process is implemented by a single master edge equipment EBi In the case of the non-hierarchical distributed mode it is implemented by all EBi edging equipment

In all cases, a device implementing the method according to the invention comprises in connection with FIG. 5, the hardware elements conventionally found in a conventional computer, namely a processor 11, a random access memory of the RAM type. 12, a read-only memory ROM 13, means 14 for transmitting / receiving messages in the telecommunications network CN According to the invention, the means 14 for sending / receiving messages are multiple and manage as many communication channels as edge equipment to control

According to the invention, the equipment 10 is connected to a database 15, which may be internal or external, comprising at least the information relating to the state of charge of the equipment that it has obtained and advantageously the thresholds of maximum authorized load that he has allocated to them.

The read-only memory 13 constitutes a recording medium in accordance with the invention. This medium stores a computer program according to the invention, this program comprising instructions for performing the steps of the load management method described above.

With reference to FIG. 6, a first exemplary implementation of the centralized embodiment of the invention is now presented schematically. According to this example, the load management method is implemented by an EC configuration equipment of the CN core network PTF telephony platform. This EC equipment exchanges information with the edge equipments EB1 to EB4, in order to modify, configure and activate the allowed load threshold values for a given period at the access networks AN1 to AN4, so that the heart of the service is not in a situation of overload following a Flash Crowds phenomenon.

This embodiment is based on the ability, at the level of the core platform, to have the overall load rate N of the telephony system at time t, and to communicate with the edge equipment, in order to require an adjustment on their part. their ability to accept calls, so that the result of all the requests for edge equipment to the platform does not cause its saturation. This requires the establishment of communication channels between the EC equipment and the edge equipment EB1 to EB4, to allow such exchanges. The configuration equipment EC obtains at least one of these channels, a piece of information Ni relating to an edge equipment EBi. It exploits this information by implementing the method according to the invention. If necessary, it uses the communication channels to inform one of the EBj border equipment of a new maximum allowed load rate M] to be considered.

Reference is now made to FIG. 7, a second example of a centralized embodiment, according to which the charge management method is implemented by an ES configuration equipment external to the telephony system. Such equipment is, for example ES supervision equipment. According to the invention, information messages INFi and INFP respectively comprising the charge rates Ni and N are sent by the edge equipment and the core platform to the ES equipment. The sending of this type of messages may be periodic or conditioned by the occurrence of an event (eg passage of a given threshold). Such messages inform the ES equipment about the current state of charge of the equipment in question. The information provided in these messages, along with the configuration parameters configured by the service operator, allow the ES equipment to make the appropriate decisions and route the specific call processing policies to each edge device accordingly. EB1 to EB4, in the form of new authorized load rates Mj, j integer between 1 and 4.

One advantage is to take advantage of available infrastructures and not to overload service and service equipment of the telephony system.

According to one variant, the ES equipment relies on measurement information relating to the load of the equipment, in particular edge equipment, which is collected, in a known manner, by RMON probes or SNMP MIB equipment. advantage of a telecommunication network supervision system set up by the operator In this case, the ES equipment can either directly interrogate the probes that feed this system (for example using polhng SNMP) through existing communication channels, or re-exploit the information already collected by the supervisory system. In addition, a one-way communication channel is sufficient between the ES equipment and the PTF service platform and each of the edge equipments EB1 to EB4 to adjust the new authorized load threshold Mi

Advantageously, the messages exchanged between the ES or EC equipment of FIGS. 6 and 7, implementing the method according to the invention and the edge equipments EB1 to EB4 are based on the DHCP protocol.

Indeed, DHCP is mainly deployed by the operators and in particular uses to communicate to the service equipment their configuration at startup This protocol is also used regularly for the update of the configuration information

Advantageously, the implementation of the invention uses a new DHCP option to communicate updates of the authorized load threshold values to the edge equipment by the supervision equipment. Preferably, this new DHCP option , called CALL_FLOW_ACCE PTANCE (CFA), will be exploited during the start of edge equipment It is defined as follows The format of the CALL_FLOW_ACCE PTANCE option is as follows

CALL_FLOW_ACCEPTANCE

Code This field informs the code of I option The value of this field must be assigned (TBA for "To Be Assigned" in English) by IANA (Internet Assigned Numbers Aυthonty) The length of this field is 1 octet Len This field informs the length of the option value In the case of the CALL_FLOW_ACCEPTANCE option, the value of this field is variable Goes/ . This field informs the value of the authorized load Mi at the level of a border equipment EBi This value is expressed in "call rate per second" or CAPS ("CaII Attempts Per Second").

It is of course understood that the DHCP protocol is only mentioned here as an example. Other means can be considered, such as the use of service configuration methods such as those based on the principles of NETCONF

(NETwork CONFiguration) or the Common Open Policy (COPS) protocol

Service).

With reference to FIG. 8, an exemplary implementation of a distributed and hierarchical embodiment of the invention is presented. According to this example, the load management method is implemented at a level of edge equipment

EB4, either nominally or to deal with a failure of EC equipment or

ES of the telecommunications network described previously. It is therefore a hierarchical mode, in which an EB1 edge equipment plays a particular role in relation to others, that of centralizing load management decisions, of implementing the decision-making logic and of notifying the other edge equipment with new permitted load rates.

In order to carry out this distributed mode, the edge equipment EB1 must, in order to implement the method according to the invention, establish communication channels that are at least unidirectional with each of the other edge equipments EB2 to EB4 so as to form a community. border equipment in which the load information can be exchanged. Two communication modes, the point-to-point mode or the point-to-multipoint mode, can be considered for the exchange of information on the charging rates of the edge equipment:

According to a first aspect of the invention, all the messages are sent in unicast to the only recipient, either the configuration equipment for the effective load rate Ni or the edge equipment EBj for the new authorized rate Mj. One advantage is to limit the number of messages exchanged and thus optimize network resources According to a second aspect of the invention, all the notification messages (Ni) are sent in unicast and the decision messages (Mj) in multicast. One advantage is to simultaneously inform the entire community of border equipment of the decisions taken. in case of failure of the master equipment, the equipment that replaces it immediately has all the updated configuration information,

According to a third aspect of the invention, all the messages are sent in multicast. The advantage of this second variant is the standardization of the information stored by the set of SBCs, whether the notifications or the decisions

Another technique would be to rely on "peer to peer" mechanisms based on the use of DHT (Distπbuted Hash Table)

Finally, with reference to FIG. 9, an example embodiment of a distributed and non-hierarchical embodiment of the invention is presented. According to this embodiment, each of the edge equipments EB1 to EB4 implements the method. According to the invention, the set of edge equipments is given the values of the global charge rate authorizes M for the telephony system and those Mi of each of the edge equipments. These values can be retrieved in various ways configured "hard" in equipment, communicated by

DHCP during commissioning, etc. Similarly, these values can be updated by the operator by the method of his choice manual configuration, by DHCP,

NETCONF, COPS, etc

According to this embodiment of the invention, the different edge equipments dynamically exchange, in unicast or multicast mode, the values of effective charge rate Ni and allow Mi which characterize them, so that each of them has an up-to-date view of all the load information of all the edge equipments of its community. This information then allows each of the edge equipments to estimate the effective load rate of the core service platform, based on a calculation formula exploiting the actual load rates of the edge equipment and configured by the service operator.

In this way, the autonomous mode allows to face the two critical situations identified above as follows ^: 1- Following a large influx of calls, the total effective load at the level of the core platform as calculated by the Ni-based edge equipment increases significantly, approaching the maximum allowable global load threshold M defined by the operator. To prevent this threshold from being exceeded, some border devices will decrease the Mi value representing the maximum load that these edge devices are allowed to transmit to the core platform. As mentioned above, the choice of the border equipment in question can be done in different ways, but it is reasonable to consider that the equipment generating the most load is self-limiting, or even rely on a decrease in Mi distributed on the set of border equipment. Typically, the Mi values can be reduced so that the effective Ni's are limited to levels such that the inequality (1) is verified. 2- Following a large influx of calls at the level of an EBi edge equipment, the latter is confronted with the situation where Ni reaches Mi. He can then consult the Ni / Mi values of the other edge equipments in order to evaluate their effective load rate Ni. In the event that one (or more) other edge equipment has a low load rate, the edge equipment close to the overload can request this last (s) so that it decreases (nt). ) temporarily its charge rate Mi and that it itself can raise its own value Mi, insofar as the total charge found at the core platform is less than n * M. When the call flow is absorbed, the edge equipment can then inform the other edge equipment, which will readjust their Mi values while the border equipment concerned reposition its Mi value to a lower level. This operation therefore amounts to temporarily adjusting the values of the Mi in order to limit the occasional phenomena of overload at the level of the equipment of border, while ensuring that the heart platform never finds itself overloaded

The new allowed rate values can be broadcast to all community edge devices in multicast mode or sent only to their recipients in unicast mode

The autonomous mode, relying on a communication channel between the edge equipments, is therefore able to manage the two crisis situations identified just as the centralized mode allowed.

Claims

A method of load management of a system for implementing an application service in a telecommunications network, said system comprising a plurality of equipment including edge equipment able to relay session establishment messages. applications between system equipment and service access networks, characterized in that said method comprises the following steps:

- obtaining an effective load rate of at least one equipment of the system;

Load management method according to claim 1, characterized in that the allowable load factor for an edge equipment is an acceptance rate of received signaling requests.

3. Load management method according to claim 1, characterized in that the effective load ratio obtained relates to a plurality of equipment forming a centralized platform.

A charge management method according to claim 1, characterized in that the effective charge rate obtained is for an edge equipment and that the step of calculating an authorized rate evaluates the possibility of assigning said equipment to border a new allowable rate that is greater than the current allowed rate and less than a maximum capacity of that border equipment

5. Load management method according to claim 4, characterized in that it further comprises a step of obtaining an effective load rate of a plurality of equipment forming a centralized platform and in that the calculation stage is implemented if the effective load rate of the centralized platform is less than an authorized load rate for said platform

6. Load management method according to claim 1, characterized in that includes a step of sending the new authorized rate to said at least one border equipment

7. Load management method according to claim 3, characterized in that the step of obtaining an effective load rate of the centralized platform comprises a step of recovering said load rate from a supervision equipment.

8 load management method according to claim 5, characterized in that the step of obtaining the effective load rate of the platform includes a step of estimating the effective load rate of the platform from the effective load rate said at least one edge equipment

9. A charge management method according to claim 4, characterized in that it further comprises a step of obtaining a permitted charge rate of said edge equipment.

10. A method for regulating the loading of an edge equipment of an implementation system of an application service in an IP telecommunications network, said system comprising a plurality of devices among which edge equipment capable of exchanging application session establishment requests with service access networks, said method comprising the following steps: sending an effective load rate to a configuration equipment of said system; receiving a new authorized charging rate from said configuration equipment, and

regulating a processing of signaling requests received by said edge equipment according to the new authorized charging rate

11. Load control method according to claim 10 characterized in that said sending means are implemented when said effective load rate reaches the authorized load rate.

12. The charge control method as claimed in claim 10, characterized in that said sending means are implemented according to a predefined sending rate. The equipment for configuring a system for implementing an application service in a telecommunications network, said system comprising a plurality of devices, among which edge equipment able to exchange application session establishment messages with service access networks, characterized in that said configuration equipment comprises the following means: effective load of at least one equipment in the system,

- if the effective load rate of said at least one equipment reaches an authorized load rate for this equipment, calculating a new load factor allowed for at least one of the edge equipment, the actual load rate of said at least one equipment being a function of said authorized load rate of said at least one edge equipment.

14. Edge equipment of a system for implementing an application service over a telecommunications network, said system comprising a plurality of equipment including edge equipment able to exchange application session establishment messages with service access networks, characterized in that it comprises the following means:

sending an effective load rate to a configuration equipment of said system;

receiving a new authorized load rate from said configuration equipment; and

regulating a processing of signaling requests received by said edge equipment according to the new authorized charging rate,

15. Edge equipment according to claim 14, characterized in that it is a configuration equipment comprising the following means:

obtaining an effective charge rate of at least one IP telephony system equipment; - if the effective load rate of said at least one equipment reaches an authorized load rate of said at least one equipment, calculating a new load factor allowed for at least one of the edge equipment, the effective load rate of said at least one equipment being a function of said authorized load ratio for said at least one edge equipment.

16 System for implementing an application service in a network of IP telecommunications, said system comprising a plurality of devices among which edge equipment able to exchange application session establishment messages with service access networks characterized in that it comprises at least one configuration equipment comprising the following means

obtaining an effective load rate of at least one equipment of the system,

- if the effective load rate of said at least one equipment reaches an authorized load rate for this equipment, calculating a new load factor allowed for at least one of the edge equipment, the effective load rate of said at least one equipment being a function of said authorized load ratio for said at least one edge equipment, and in that the edge equipment comprises the following means - sending an effective load rate to a configuration equipment of said system,

receiving a new authorized load rate from said configuration equipment, and

17. Computer program product downloadable from a communication network and / or stored on a computer readable medium and / or executable by a microprocessor, characterized in that it comprises program code instructions for the execution of the method load management system according to claim 1

18. Computer program product downloadable from a communication network and / or stored on a computer readable medium and / or executable by a microprocessor, characterized in that it comprises program code instructions for the execution of the method load control device according to claim 10