GB2423436A

GB2423436A - Media proxy load balance management

Info

Publication number: GB2423436A
Application number: GB0503409A
Authority: GB
Inventors: Guy Badman; Martin Wakley
Original assignee: Nortel Networks Ltd
Current assignee: Nortel Networks Ltd
Priority date: 2005-02-18
Filing date: 2005-02-18
Publication date: 2006-08-23
Also published as: GB0503409D0

Abstract

The present invention relates to methods of, computer software for and apparatus for improved management of media proxy (MP) resources. More particularly, but not exclusively, the present invention enables, in a carrier hosted solution, intelligent selection of media proxies to balance load across the carrier network. According to the present invention, MPs signal information back to the call agents that control them. Preferably, the signalling is added to keep alive messages such as Media Gateway Control Protocol (MGCP) Restart in Progress (RSIP) messages.

Description

IMPROVED MEDIA PROXY RESOURCE MANAGEMENT

Field of the Invention

The present invention relates to methods of, computer software for and apparatus for improved management of media proxy resources. More particularly, but not exclusively, the present invention enables, in a carrier hosted solution, intelligent selection of media proxies to balance load across the carrier network.

Backcjround In a carrier hosted Voice over Internet Protocol (VolP) solution, an enterprise or residential network will include terminal devices incorporating Media Gateways (MG5) with which users may make and receive voice or other media calls. Typically, Network Address Translators (NATs) and/or firewalls are deployed to provide lP address resolution and security between the customer IP networks and the carrier IP network. Additionally, the carrier network will employ large trunking MGs to interface to conventional circuit-switched networks such as the Public Switched Telephone Network (PSTN). Calls between VoIP terminals and legacy telephone terminals in the PSTN will be routed through these trunking MGs.

However, the presence of NATs and/or firewalls separating the customer lP networks and the carrier IP network creates additional call establishment issues for the carrier. In particular, wherever a call is established to one or more end users whose MGs reside behind a NAT and/or firewall, a Media Proxy (MP) must be used in the carrier IP network to act as a recognisable point of contact for other call participants whose MGs cannot communicate directly with MGs residing behind the NAT as a result of IP address translation. MPs act as a "bridge" with which all call participants may communicate whether their MGs reside behind a NAT and/or firewall or not. The function of the MP is simply to act as a point of contact addressable to any call participant and to forward IP media streams from one call participant to others. MPs are sometimes referred to as Media Portals or Real Time Protocol (RTP) portals.

With large carrier networks servicing many thousands of customer networks each with large numbers of VoIP terminals, it can be seen that a large number of MPs will typically be required and that the loading of these MPs will fluctuate rapidly as calls are set up, modified and removed. Currently, call agents in call servers (CSs) in the carrier network use a "round robin" mechanism for selecting a MP to use for a given call which requires it. In other words, each call agent selects in turn a successive MP from a pool of MP it is associated with.

The control mechanism for requesting a MP to act in respect of a given call uses a modified version of the Media Gateway Control Protocol (MGCP). Using this protocol, a MP is requested to accept a call by a call agent or by a Media Gateway Controller (MGC) acting as a proxy for the call agent. The MP may accept the call or refuse, for example if it is overloaded.

One problem with the current mechanism occurs when a number of separate call agents are arranged to be associated with a shared pool of MP5.

This is desirable for resiliency and also for network efficiency and scalability, since MPs should ideally be as central as possible, geographically or topologically, between the participating MGs. However, in such arrangements, imbalances in the loading of MPs can easily occur. Different communities of interest in the customer networks (possibly at different times of day etc) can result in different levels of call activity at different call agents thereby increasing the load on MPs associated with those call agents, but not on other MPs. Other call agents sharing those MPs with increased load will not be aware of the increased load and cannot compensate for it. Imbalances in loading MPs are undesirable. Firstly, it is inefficient use of the carrier network's resources which is costly. Secondly, it can lead to increased signalling in the carrier network, due to MPs being overloaded and rejecting requests from call agents resulting in the call agents having to repeat the process until a MP with the capacity to service the request is found. This is again inefficient use of resources and can lead to network scalability problems. Thirdly, it can, in worst case scenarios, lead to denying call requests.

An object of the present invention is too overcome or ameliorate the problems described above.

Summary of the Invention

According to one aspect of the present invention, there is provided a method of determining whether to select a media proxy for use in a media call established at least partially over a packet-switched network, the method comprising: a first call agent controlling a media proxy in the packet-switched network to proxy call traffic in respect of a first media call; the media proxy generating data indicating its current loading, said loading being at least partially caused by said media proxy proxying call traffic in respect of said first media call; the media proxy sending the data to a second call agent; and the second call agent determining whether to select the media proxy for use in a second media call in dependence on said data.

Advantageously, the second call agent is able to make an informed decision whether or not to select the media proxy for use in a second media call taking into account the loading of the media proxy as a result of the first media call even though the first media call is controlled by a separate call agent. Thus, the method enables improved balancing of loads across many media proxies and increased efficiency, scalability and quality of service as a result.

In one embodiment, said media proxy sends data indicating its current loading to each call agent which is capable of controlling it.

In one embodiment, said media proxy sends said data in a keep alive message.

Thus, the keep alive mechanism is re-used for load balancing without requiring significant extra signalling or processing resources.

In one embodiment, said media proxy sends said data in a Media Gateway Control Protocol (MGCP) message. In one embodiment, said MGCP message is a Restart In Progress (RSIP) message.

In one embodiment, the media proxy periodically generates data indicating its current loading and sends the data to the second call agent. Thus, the second call agent is kept updated with current loading of the media proxy.

In one embodiment, the media proxy sends the data to the second call agent in response to an event. In one embodiment, the event is the media proxy exceeding a predetermined loading threshold. Thus, signalling and processing is reduced when the media proxy is not above the predetermined loading threshold.

According to another aspect of the present invention, there is provided a media proxy for use in a media call established at least partially over a packet-switched network, the media proxy comprising: means for receiving requests from a plurality of call agents whereby each of said call agents is able to control said media proxy to proxy call traftic; means for generating data indicating its current loading; means for sending the data to said plurality of call agents.

According to another aspect of the present invention, there is provided a call agent for use in a packet-switched network, the call agent comprising: -5..

means for controlling a media proxy in the packet-switched network to proxy call traffic in respect of a media call; means for receiving data indicating the current loading of the media proxy, said loading being at least partially a result of said media proxy proxying call traffic in respect of a media call under the control of another call agent; means for determining whether to select the media proxy for use in a second media call in dependence on said data.

According to another aspect of the present invention, there is provided a computer program for use in a call agent, the computer program comprising: code for controlling a media proxy in a packet-switched network to proxy call traffic in respect of a media call; code for receiving data indicating the current loading of the media proxy, said loading being at least partially a result of said media proxy proxying call traffic in respect of a media call under the control of another call agent; code for determining whether to select the media proxy for use in a second media call in dependence on said data.

According to another aspect of the present invention, there is provided a computer program for use in a media proxy, the computer program comprising: code for receiving requests from a plurality of call agents whereby each of said call agents is able to control said media proxy to proxy call traffic; code for generating data indicating its current loading; code for sending the data to said plurality of call agents.

Other aspects of the present invention provided software for performing the above methods, whether in the form of a computer program per se or recorded on a data carrier.

Further advantages and aspects of the present invention will be apparent

from the detailed description provided below.

There now follows, by way of example, a detailed description of the present invention in which:

Brief Description of Drawings

Figure 1 is a schematic diagram showing a typical carrier-hosted V0IP

network arrangement according to the prior art;

Detailed Description of the Invention

Figure 1 shows a carrier-hosted VoIP network arrangement in which a carrier IP network 10 is connected to an lP network 12 via a number of NATs and/or firewalls 14. The NAT/firewalls 14 separate a number of customer IP networks and the carrier P network 10. The present invention applies equally to NAT only implementations, firewall only implementations and NAT plus firewall implementations. In NAT implementations, NAT/firewall 14 may comprise any type of NAT including full cone, restricted cone, port restricted cone, symmetric NATs and may comprise a Network Address and Port Translator (NAPT) or Network Address Translator and Protocol Translator (NAT-PT). Carrier A plurality of devices 16, 18 in network 12 provide users with VoIP services. For example, devices 16, 18 may comprise IP telephony, facsimile, or video conferencing equipment whether implemented as specific hardware devices or as software clients running on generic computing devices.

Furthermore, devices 16, 18 may comprise IP Private Branch Exchanges (PBXs) serving multiple user terminals, cable modem equipment and so on.

Carrier IP network 10 is also connected to PSTN 24 via a trunking MG 20.

A plurality of conventional circuit-switched telephones 22 provide users in legacy voice, fax and data etc. services in the PSTN 24.

Two media streams 26 and 30 are shown representing two exemplary calls. Media stream 26 serves a call between one of conventional telephones 22 in PSTN 24 and one of MGs 16 in IP network 12. As can be seen, media stream 26 passes through one of NATs 14 and thus a MP 28 is used to forward RTP packets between, and to act as the recognisable point of contact for, the call endpoints. MG 20 transcodes between RTP packets on the carrier network side and circuit-switched PCM signals on the PSTN side. Media stream 30 serves a call between one of MGs 16 and one of MGs 18 in IP network 12 and passes through two NATs 14. This would normally occur when two private customer networks are provided V0IP services by carrier network 10. Again, a MP 32 is used to forward RTP packets between, and to act as the recognisable point of contact for, the call endpoints. Other call examples are possible involving two or more end user terminals, point to point or point to multi-point, unidirectional or bidirectional call traffic and any number of NATs. Typically, however, only one MP will be used for a single call hosted by carrier network 10.

Carrier network 10 also includes a number of call servers 34 which control the establishment, management and tearing down of calls to devices 16, 18 and telephones 22 according to a call control protocol. For example, the Session Initiation Protocol (SIP) may be used as the call control protocol in which case call server 34 is a SIP server and devices 16, 18 each comprise a SIP User Agent (UA). Other call control protocols may be used such as MGCP, H.323, H.248 (Megaco) and Network Call Signalling (NCS) may be used. Each CS 34 will typically execute a plurality of call agents. Carrier network 10 also includes a plurality of Media Gateway Controllers (MGC5) 36 which control MGs 16, 18, 20.

using the Media Gateway Control Protocol (MGCP). In response to a request to establish a call, a call agent in one of CSs 34 instructs appropriate MGs and MPs to establish communications channels and reserve resources to enable the call to proceed. This instruction is performed via one or ones of MGCs 36. In other arrangements, CSs 34 and MGCs 36 may be co-located or even integrated functions.

According to the present invention, MPs signal information back to the call agents that control them. In particular, MPs signal back information indicating their current load. Figure 2 shows carrier IP network 10 with two CSs 34a and 34b, each having a number of call agents, connected MP5 28a, 28b and 32 via MGCs 36. In the example, each CS 34a and 34b is capable of controlling each MP 28a, 28b and 32. However, because call agents are logically separate, each call agent will only have a partial "view" of the state of each of the MPs. Other call agents on the same CS or on other CSs may be using the resources of the same MPs. To provide a more global view of resource capacity on MPs 28a, 28b and 32, each MP signals its current loading to all of the call agents that are able to control it. In Figure 2, the signalling from MPs 28a and 28b is shown as dotted lines 40a, 40b flowing from each of MPs 28a and 28b to each of CSs 34a and 34b. By receiving signalling from each MP it controls, a call agent in one of CS5 34a and 34b is able to "see" the loading on MPs caused not only by itself, but by other call agents on the same CS or on other CSs and is able to select MPs to serve future calls thereby to balance load across all MPs it controls.

To maintain a signalling association between call agents and MPs, a "keep alive" message is conventionally sent from each of MPs 28a, 28b and 32 to the call agents they serve. For example, with MGCP as the MP control protocol, the Restart in Progress (RSIP) message is conventionally used as the keep alive message. Such a message is periodically sent to inform the call agent that the MP is still functioning. For example, RFC 3435 describes an example RSIP message informing the call agent that all the gateway's endpoints are being placed in-service in 0 seconds, i.e., they are currently in service: RSIP 1204 *@rgw_2567 whatever net MGCP 1.0 RN: restart RD: 0 In one embodiment of the present invention, the RSIP message is used to carry additional data indicating the current loading of MP.

RSIP 1204 *@rgw_2567 whatever net MGCP 1.0 RN: restart RI): 0 LI) : 87% Other MGCP commands such as the AuditEndpoint (AUEP) command may be used as keep alive messages. In the case of AUEP, which is sent by the call agent to MPs, the indicating the current loading of MP is sent in a response to the AUEP command. In general, any MGCP or other message may be used to carry data indicating the current loading of MP, whether initiated by the MP itself or as a response to a message initiated by the call agent.

Messages may be sent periodically, as is the case with keep alive messages, or in response to events or both. For example, a message may be sent by a MP to its call agents in the event that its load exceeds a predetermined threshold.

In further embodiments of the present invention, carrier network 10 also includes a Network Management System (NMS) arranged to monitor the performance and operational condition of equipment in the network. For example, under the terms of a Service Level Agreement between the carrier and the customer, the carrier may be required to record network availability and report failures within a prescribed time period. In this case, in addition to data indicating the current load, data indicating the performance and/or operational condition of MPs may be signalled to call agents in the same signalling messages. This data may then be forwarded by the call agents to the NMS.

The loading (and performance/operational condition) data may be generated in plain text, a mark up language such as XML, or a succinct binary representation. - 10-

In general, the operational condition information may be inserted in the payload or header portions of keep alive messages.

It will be appreciated that the present invention has application to network arrangements other than carrier hosted VoIP arrangements. In general, the present invention has application to any network arrangement where media proxy resources are used to establish communications between two or more call parties where one or more call parties reside behind a NAT/firewall.

Claims

Claims: 1. A method of determining whether to select a media proxy for use

in a media call established at least partially over a packet-switched network, the method comprising: a first call agent controlling a media proxy in the packet-switched network to proxy call traffic in respect of a first media call; the media proxy generating data indicating its current loading, said loading being at least partially caused by said media proxy proxying call traffic in respect of said first media call; the media proxy sending the data to a second call agent; and the second call agent determining whether to select the media proxy for use in a second media call in dependence on said data.
2. A method according to claim 1, wherein said media proxy sends data indicating its current loading to each call agent which is capable of controlling it.
3. A method according to claim 1, wherein said media proxy sends said data in a keep alive message.
4. A method according to claim 1, wherein said media proxy sends said data in a Media Gateway Control Protocol (MGCP) message.
5. A method according to claim 4, wherein said MGCP message is a Restart In Progress (RSIP) message.
6. A method according to claim 1, wherein the media proxy periodically generates data indicating its current loading and sends the data to the second call agent. - 12-
7. A method according to claim 1, wherein the media proxy sends the data to the second call agent in response to an event.
8. A method according to claim 7, wherein the event is the media proxy exceeding a predetermined loading threshold.
9. A media proxy for use in a media call established at least partially over a packet-switched network, the media proxy comprising: means for receiving requests from a plurality of call agents whereby each of said call agents is able to control said media proxy to proxy call traffic; means for generating data indicating its current loading; means for sending the data to said plurality of call agents.
10. A media proxy according to claim 9, wherein said data is sent in a keep alive message.
11. A media proxy according to claim 9, wherein said data is sent in a Media Gateway Control Protocol (MGCP) message.
12. A media proxy according to claim 11, wherein said MGCP message is a Restart In Progress (RSIP) message.
13. A media proxy according to claim 9, wherein the media proxy is arranged to periodically generate and send said data.
14. A media proxy according to claim 9, wherein the media proxy is arranged to send said data in response to an event.
15. A media proxy according to claim 14, wherein the event is the media proxy exceeding a predetermined loading threshold.
16. A call agent for use in a packet-switched network, the call agent comprising: means for controlling a media proxy in the packet-switched network to proxy call traffic in respect of a media call; means for receiving data indicating the current loading of the media proxy, said loading being at least partially a result of said media proxy proxying call traffic in respect of a media call under the control of another call agent; means for determining whether to select the media proxy for use in a second media call in dependence on said data.
17. A call agent according to claim 16, wherein said data indicating the current loading of the media proxy is received in a keep alive message.
18. A call agent according to claim 16, wherein said data indicating the current loading of the media proxy is received in a Media Gateway Control Protocol (MGCP) message.
19. A call agent according to claim 18, wherein said MGCP message is a Restart In Progress (RSIP) message.
20. A computer program for use in a call agent, the computer program comprising: code for controlling a media proxy in a packet-switched network to proxy call traffic in respect of a media call; code for receiving data indicating the current loading of the media proxy, said loading being at least partially a result of said media proxy proxying call traffic in respect of a media call under the control of another call agent; code for determining whether to select the media proxy for use in a second media call in dependence on said data. - 14
21. A computer program for use in a media proxy, the computer program corn prising: code for receiving requests from a plurality of call agents whereby each of said call agents is able to control said media proxy to proxy call traffic; code for generating data indicating its current loading; code for sending the data to said plurality of call agents.