Title of Invention

"VERIFICATION OF THE AVAILABILITY OF A SERVER"

Abstract According to the invention, an availability request is transmitted by a client to a server in order to verify whether a server is available. A response to said availability request is transmitted by the server to the client by means of a confirmation message if the server is available. The client transmits a message regarding the availability of the server to other clients who then prevent the transmission of an availability request to the server at least for a predefined period of time.
Full Text The present invention relates to a method for verifying the availability of a server, a control program and a client for a network providing connectionless services.
Reliable verification of the availability of a server is of great significance in particular in the case of loosely connected client-server relations in packet-oriented data networks. If a client detects early enough that a server is overloaded or down, then it is possible to implement countermeasures in good time, such as, for example, the search for an alternative server or the creation of warning notices. Methods for verifying the availability of a server are applied, which methods also include keepalive tests described in the H.323 standard, as on 11/2000, chapter 7.2.2, if there is no permanent communication relationship between client and server and if, however, such a relationship must exist and be without fault is the basis for a desired functionality, for example, Internet telephony. In packet-oriented networks keepalive tests are used, for example, to simulate to the communication subscribers a quasi line-switching character with respect to mutual availability.
With conventional keepalive tests, a client sends availability requests to a selected server in cyclical intervals of time. If the server transmits a response to the availability request within a predefinable period of time, the server is deemed available and thus the communication relationship active. Of disadvantage in this approach is that the server is generally put under a severe load by transmitting a response to all the client requests.
The object underlying the invention is thus to specify a method for verifying the availability of a server, which method enables the load placed on the server by the availability requests directed at it to be minimized, as well as to specify technical implementations suitable for carrying out said method.
This object is achieved according to the invention by a method with the features specified in claim 1, a control program with the features specified in claim 7 and a client with the features given in claim 8. Advantageous developments of the present invention are set down in the dependent claims.
An essential aspect of the present invention consists in that a client, which has received a confirmation message from a server in response to an availability request directed to said server, transmits a message regarding the availability of the server to other clients. Whereupon the predefinable other clients prevent the transmission of an availability request to the server at least for a predefinable period of time. In this way, it is possible to considerably reduce the load placed on the server by availability requests, without this being at the cost of, for example, a server failure or a server overload no longer being detected in good time.
The present invention is explained in greater detail below in an exemplary embodiment with reference to a drawing, in which
Figure 1 shows an application environment of the present invention with a large number of clients and servers in a packet-oriented network with several subnetworks,
Figure 2 shows a flow chart for a method for verifying the availability of a server,
Figure 3 a flow chart for, compared with Figure 2, a modified method in which, in addition, the availability of clients is checked by the server.
Figure 1 displays a communication system that comprises a packet-oriented network with several subnetworks 101, 110, 120, 130, server 102 to 104 and several clients 111 to 115, 121 to 123, 131 to 13 3. The clients 111 to 115, 121 to 123, 131 to 133 use the services offered by the servers 102 to 104, for example, Internet telephony.
The subnetworks 110, 120, 130 are local subnetworks, each of which has a gatekeeper. The gatekeepers are used mainly for access control and address translation services. Servers in the packet-oriented network can assume this functionality of the gatekeeper.
The clients 111 to 115, 121 to 123, 131 to 133 first register with the respective server in order to use the services offered by a server 102 to 104. Thereby, a period of time is specified within which a registration is treated as valid. The sequence of a registration process is represented in model form and without any limitation of universal validity for client 111.
Client 111 has a central processing unit (CPU), a random access memory (RAM), a hard drive for non-volatile storage of data (HD), a network adaptor (Tx/Rx) and a timing unit (timer). The remaining clients 112 to 115, 121 to 123, 131 to 133 represented in Figure 1 can also have such a structure and such a functionality.
In order to register with one of the servers 102 to 104, the client 111 transmits a message with a request to register RRQ with a server that provides a desired service. In the message with the request to register RRQ, there is a period of time specified within which period the registration of client 111

with the server selected must be valid. Outside this period of time, the registration of the client is considered invalid.
If it is available the selected server responds to the registration request with a message containing registration confirmation RCF. In the message with the registration confirmation RCF, a period of validity is also specified for the registration. The period of validity confirmed by the selected server either corresponds to the period of time specified by client 111 or is shorter than said period of time.
Within the period of validity for the registration, client 111 can direct availability requests to the selected server. Availability requests of this type, which are also called keepalive tests, are likewise transmitted to the selected server in the form of a message with a registration request RRQ. A message with a registration request RRQ sent in the course of a keepalive test has, compared to the usual messages with a registration request, a set keepalive bit.
In accordance with the ITU-T recommendations H.225.0, as on 11/2000, chapter 7.9, a shortened message with a registration request is adequate for a "lightweight registration". Thereby the message with a registration request only has information about the client's registration and status transport address (RAS address), a set keepalive bit, an end point indicator for the client, a server or gatekeeper identifier, tokens for operations to be carried out and a period of validity for the registration.
A message with request to register RRQ, which is transmitted to the selected server during a keepalive tests, can also be used to extend the period of validity of the registration. However, this only applies if the message is received by the selected server within the period of validity of the registration. Therefore, messages und processing delays should be taken into consideration when a client is notifying the end of a period of validity. After the registration validity period has ended it is no longer sufficient to transmit a shortened message with a registration request as described above in order to renew the registration.
If the message with the registration confirmation RCF does not have any information as to the period of validity of the registration, this will be interpreted as indicating that the selected server does not support any keepalive mechanisms for availability requests. If clients receive such a message without information about a period of validity, they should prevent availability requests being transmitted to the respective server.

When the validity of a registration has expired, the selected server can send a message to the client in question with a corresponding indication in order to inform said client that the period of validity has expired. This is of particular advantage when there is a loss of synchronization between the control units at the client on the one hand and at the server on the other.
If, after the period of validity of the registration has ended, client 111 transmits an above described shortened message with registration request RRQ and set keepalive bit to the selected server, then the server will respond with a message with registration rejection RRJ. A reason for rejection can also be included in such a message.
In order to avoid loading a server 102 to 104 unnecessarily with availability requests from clients 111 to 115, 121 to 123, 131 to 133, a client that has received a positive or a negative response to an availability request, uses a multicast message MCM to inform all the clients within the same subnetwork 110, 120, 130 of the availability of the respective server. In addition, a limitation on the same subnetwork allows problems with respect to generally unsolvable addresses to be avoided. In particular, unnecessary loading of the network with non-addressable messages is prevented.
The tlow chart represented in Figure 2 serves to further illustrate a method for verifying the availability of a server in a packet-oriented communications network. The starting point of the method is the start of a timer or of a counter of a timing unit of a client (step 201). Using the timer, times are set at which the client's availability request are to be transmitted to a selected server.
After the timer has started, the client monitors the reception of the multicast messages MCM, which provide statements as to the availability of a server (step 202). If the client receives such a multicast message, then the timer is reset (step 208). This thus prevents the transmission of availability requests to the selected server for a predefmable period of time.
Provided that no new multicast message was received, in addition a check is made whether the timer has run out and thus a time arrived at for delivering a new availability request (step 203). If the timer has not yet run out, then the reception of multicast messages and the progress of the timer continues to be checked. If the timer has run out, an availability request is sent to the selected server (step 204).
In slep 205, the availability or unavailability of the selected server is determined at least implicitly. If the selected server is available, then said server responds to the availability request by sending a confirmation message to the requesting client (step 206). After having received the confirmation
message, the requesting client informs all other clients within the same subnetwork about the availability of the selected server by means of a multicast message (step 207). The other clients who have received the multicast message, hereupon for their part prevent the transmission of availability requests to the selected server for a predefinable period of time.
As, when the selected server fails or is overloaded, no confirmation message regarding the availability of said server is sent, a check is made to determine whether the selected server is at least able to respond to the availability request with an unavailability message (step 209). If the selected server were overloaded, this would, for example, still be possible. By contrast, if the selected server fails, one can no longer expect an unavailability message to be sent. Therefore, a check is made to determine whether a time limit for the reception of a confirmation message or of an unavailability message has been reached (step 211). If this time limit is exceeded then the requesting client transmits a multicast message regarding the unavailability of the selected server to all clients within the same subnetwork (step 210). The same procedure is followed if the requesting client receives an unavailability message within the time limit allowed for a response to the availability request.
The method for verifying the availability of a server described is preferably implemented by means of a control program which can be loaded in a random access memory of a client and has at least one code section, on the implementation of which control program, the steps described above proceed to run - transmission ol" the availability request, monitoring of the receipt of a confirmation message, transmission of a message regarding the availability to other clients, monitoring the receipt of messages of other clients and if necessary preventing the sending of availability requests. By means of the control program, a client for a communications network that provides connectionless services is implemented, which client has a device for transmitting availability requests, a device for monitoring the receipt of confirmation messages, a device for transmitting messages regarding the availability of servers to other clients as well as a device for monitoring the receipt of messages of other clients and for preventing the transmission of availability requests.
The following observations serve to clarify the advantages of the described method for checking the availability of a server compared with conventional keepalive tests used up to now. According to the conventional keepalive tests used up to now a server with n=1000 clients sending availability requests and a request rate of a=3 requests per minute and client, and assuming such requests were equally distributed in time, would have to respond to an availability request every

60s 60s
[|t, (a = 3, n = 1000) - a . n = 3,1000 = 20ms]]..
Thus, the server would be busy answering an availability request in real time on average every 20 ms. This would result in a considerable load being placed on the server by availability requests, which would restrict it greatly in realizing the tasks which it was actually intended for. The first option to present itself for solving this problem would be to make a large reduction in the clients' request rates per minute, which would, however, result in the affected clients possibly determining the loss of the availability of the server far too late.
To assess one of the advantages that can be achieved with the method proposed here, it is first assumed that n clients are distributed on subnetworks. It is further assumed that from each of these subnetworks at least one client will try to query the availability of the server. Each of these clients would then inform the remaining clients within its subnetwork by means of a multicast message regarding the availability of the server. Under ideal conditions, the number of availability requests to be processed by the server would correspond to the number s of subnetworks. In addition, it must also be taken into consideration that the multicast messages in the subnetworks could have a loss rate v between 0 and 100 percent. Thus the number vc of clients per subnetwork, which are not reached by a multicast message and hence themselves send an availability request to the server depending on the loss rate v of the multicast messages:
(Figure Removed)
The total number vcall of all the clients who, under this condition, send an availability request to the server, can thus be estimated as follows:
(Figure Removed)
Assuming the above, the period of time lr between two availability requests sent to the server can be calculated as follows:

(Figure Removed)
Using a numerical example, the load placed on the server by availability requests in the conventional keepalive tests up to now is compared with the load placed on the server in accordance with the method proposed here. With a request rate of a=3 availability requests per minute and n=1000 clients, the period of time t, between two availability requests is:
(Figure Removed)
Typical values for the loss rate v are clearly below 100 percent. With the above values for the request rate a, the number n of the clients and s=10 subnetworks, produces, in dependence on the loss rate v, the following efficiency increases deff in percent in relation to the conventional keepalive tests with a period of time tr = 20 ms between two availability requests:

(table Removed)
With a loss rate v = 20 % the mean time tr between two availability requests to be processed by the server is approximately 96 ms. This corresponds to an efficiency increase defT= 380 % compared with conventional keepalive tests up to now.
All in all, one can determine that, with the method proposed for checking the availability of the server, the load placed on the server by responding to availability requests clearly lessens, without this
resulting in a significant increase in the time until the unavailability of the server is known to the clients. Moreover, in order to implement the method proposed here, changes are only necessary on the client side. The previous procedure for processing availability requests can be kept on the server side and thus does not require any further expenditure.
In the method represented in Figure 3, a server also checks the availability of clients. This means that a server is also informed about the availability of clients who are only informed about the availability of the server by another client per multicast message and who do not themselves send keepalive requests directly to the server. First each client starts a first timer (timer 1) (step 301) and checks whether a multicast collective request has been received (step 302). A multicast collective request can be sent by a client as a sign that this client would like to carry out a collective keepalive test with a server, and that this client needs the keepalive data of other clients for this.
The status of the first timer is checked continuously by the respective client (step 303). If the first timer has not yet run out and no multicast collective request has been received, then the receipt of a multicast collective request continues to be monitored. If the first timer has run out and no multicast collective request has been received, then the respective client takes on the task of sending a multicast collective request to a selected server itself. To this end the respective requesting client starts a second timer (timer 2) (step 304) and transmits a multicast collective request to predefinable other clients (step 305). Subsequently, the respective requesting client checks whether the multicast collective request was responded to by the predefinable other clients (step 306), and whether the second timer has run out (step 307). If the second timer has not yet run out and no multicast collective request has been received, then the receipt of a response to the multicast collective request continues to be monitored.
After the second timer has run out, the respective requesting client starts a third timer (timer 3) (step 308) and sends a collective availability request to the selected server (step 309). The collective availability request includes data of the respective requesting client and of those predefinable other clients who responded to the multicast collective request of the respective requesting client until the second timer ran out. Subsequently the respective requesting client checks whether the selected server has signaled availability (step 310), and whether the third timer has run out (step 311). If the third timer has not yet run out and no response has been received to the collective availability request, then the receipt of a response to the collective availability request continues to be monitored.
If the selected server has signaled its availability to the respective requesting client, then the respective requesting client transmits a positive multicast availability message to the predefmable further clients (step 312). If the selected server has signaled its unavailability or has not sent any response to the collective availability request by the time the third timer has run out, then the respective requesting client transmits a negative multicast availability message to the predefmable other clients (step 313).
Predefmable other clients who have received a multicast collective request in accordance with the check in step 302, transmit their keepalive data to the respective requesting client (step 314), the receipt of which data the respective requesting client checks in accordance with 306. Subsequently, the predefmable other clients start a fourth timer (timer 4) (step 315) and check whether a positive or negative multicast availability message of the respective requesting client was received (step 316), and whether the fourth timer has run out (step 317). On receipt of a multicast availability message of the respective requesting client, the first timer is reset (step 318), so that the predefinable other clients do not send any availability requests to the selected server for a period of time determined by the first tinier running out.
As long as the fourth timer has not run out, the predefinable other clients wait for a multicast availability message of the respective requesting client. If this does not occur before the fourth timer runs out, the respective client then itself assumes a role as the placer of a collective availability request in accordance with steps 304 to 313, provided the first timer has run out or no further multicast collective request has been received before said first timer has run out. The application of the present invention is not limited to the exemplary embodiment described here.



WE CLAIM;
1. Method for verifying the availability of a server, in which
- an availability request is transmitted by a client to a server,
- a response to said availability request is transmitted by the server to the client by means of a
confirmation message if the server is available,
characterized in that the client transmits a message regarding the availability of the server to other clients who then prevent the transmission of an availability request to the server at least for a predefmable period of time.
2. Method according to claim 1,
characterized in that data is transmitted between the server and the clients by means of connectionless switching control.
3. Method according to one of the claims 1 or 2,
characterized in that the message regarding the availability of the server is transmitted to the predefmable other clients using multicast.
4. Method according to one of the claims 1 to 3,
characterized in that the client informs only predefinable other clients within the same subnetwork about the availability of the server.
5. Method according to one of the claims 1 to 4,
characterized in that the client executes availability requests at times predefinable by a timer control.
6. Method according to claim 5,
characterized in that a timer of a time control, which control is assigned to a predefinable other client and instigates availability requests, is reset to a predefinable value when a message regarding the availability of the server is received.
7. Control program that can be loaded into a random access memory of a client and has at least one
code section, on the implementation of which control program
- an availability request is transmitted by the client to a server,
- a receipt of a confirmation message responding to the availability request if the server is available is
monitored,
- a message regarding the availability of the server is transmitted to other clients,
- a receipt of a message of a predefinable other client regarding the availability of the server is
monitored and the transmission of an availability request to the server is prevented at least for a predefinable period of time when such a message is received, if the control program in the client runs out.
8. Client for a communications network providing connectionless services with
- a device for transmitting an availability request to a server,
- a device for monitoring the receipt of a confirmation message responding to the availability request
if the server is available,
- a device for transmitting a message regarding the availability of the server to other clients,
- a device for monitoring the receipt of a message of a predefinable other client regarding the
availability of the server and for preventing the transmission of an availability request to the server at least for a predefinable period of time when such a message is received.
9. Method for verifying the availability of a server, substantially as hereinbefore described with
reference to the accompanying drawings.



Documents:

1518-delnp-2006-Abstract-(03-08-2012).pdf

1518-delnp-2006-Abstract-(04-03-2015).pdf

1518-delnp-2006-abstract.pdf

1518-DELNP-2006-Assignment-(02-11-2012).pdf

1518-delnp-2006-Claims-(03-08-2012).pdf

1518-delnp-2006-Claims-(04-03-2015).pdf

1518-delnp-2006-claims.pdf

1518-delnp-2006-Correspondence Others-(03-08-2012).pdf

1518-delnp-2006-Correspondence Others-(04-03-2015).pdf

1518-delnp-2006-Correspondence Others-(10-07-2012).pdf

1518-delnp-2006-correspondence others.pdf

1518-DELNP-2006-Correspondence-Others-(02-11-2012).pdf

1518-DELNP-2006-Correspondence-others-(04-12-2009).pdf

1518-DELNP-2006-Correspondence-others-(12-05-2006).pdf

1518-DELNP-2006-Correspondence-others-(13-08-2007).pdf

1518-DELNP-2006-Correspondence-others-(29-03-2006).pdf

1518-delnp-2006-description (complete).pdf

1518-delnp-2006-Description-(Compete)-(03-08-2012).pdf

1518-delnp-2006-Drawings-(03-08-2012).pdf

1518-delnp-2006-Drawings-(04-03-2015).pdf

1518-delnp-2006-drawings.pdf

1518-delnp-2006-Form-1-(03-08-2012).pdf

1518-DELNP-2006-Form-1-(12-05-2006).pdf

1518-DELNP-2006-Form-1-(13-08-2007).pdf

1518-delnp-2006-form-1.pdf

1518-DELNP-2006-Form-13-(10-07-2012).pdf

1518-delnp-2006-form-13.pdf

1518-delnp-2006-form-18.pdf

1518-delnp-2006-Form-2-(03-08-2012).pdf

1518-delnp-2006-Form-2-(04-03-2015).pdf

1518-delnp-2006-form-2.pdf

1518-delnp-2006-form-26.pdf

1518-delnp-2006-Form-3-(03-08-2012).pdf

1518-delnp-2006-Form-3-(04-03-2015).pdf

1518-delnp-2006-form-3.pdf

1518-delnp-2006-Form-5-(03-08-2012).pdf

1518-delnp-2006-Form-5-(04-03-2015).pdf

1518-delnp-2006-form-5.pdf

1518-DELNP-2006-GPA-(02-11-2012).pdf

1518-delnp-2006-GPA-(10-07-2012).pdf

1518-delnp-2006-pct-210.pdf

1518-delnp-2006-pct-304.pdf

1518-delnp-2006-Petition-137-(03-08-2012).pdf

1518-delnp-2006-Petition-137-(04-03-2015).pdf


Patent Number 265766
Indian Patent Application Number 1518/DELNP/2006
PG Journal Number 12/2015
Publication Date 20-Mar-2015
Grant Date 13-Mar-2015
Date of Filing 21-Mar-2006
Name of Patentee SIEMENS AKTIENGESELLSCHAFT
Applicant Address WITTELSBACHERPLATZ 2, 80333 MüNCHEN, GERMANY.
Inventors:
# Inventor's Name Inventor's Address
1 CHRISTIAN SCHEERING ERFTWEG 38, 33689 BIELEFELD, GERMANY.
PCT International Classification Number H04L 29/06
PCT International Application Number PCT/EP2004/052163
PCT International Filing date 2004-09-14
PCT Conventions:
# PCT Application Number Date of Convention Priority Country
1 10345535.3 2003-09-30 Germany