DE102015212354A1 - A method for improved load balancing with respect to the provision of a network service in a computer network, system for improved load distribution with respect to the provision of a network service in a computer network, program and computer program product - Google Patents

Abstract

The invention relates to a method of improved load sharing for providing a network service in a computer network, the computer network comprising a plurality of servers for providing the network service to a client device, the plurality of servers comprising a plurality of primary servers, the plurality of servers including at least one secondary server for failover wherein the computer network comprises a Domain Name System (DNS) server for providing Service Resource Record (SRV) records, the Service Resource Record (SRV) records having load distribution data and prioritization data, the client device communicating with the computer network to use the Network Service, the method comprising the steps of: - determining, by the client device, a first server, wherein the first server, depending on the load distribution data of the Service Resource Record (SRV) records from the plurality of Prim Providing - in the case where the first server is available to provide the network service to the client device - the network service to the client device through the first server, and - in the event that the first server fails to provide the network service to the client device and another server of the plurality of primary servers is available to provide the network service to the client device - determining, by the client device, a second server from the at least one secondary server, wherein the second server is determined in dependence on the load distribution data of the Service Resource Record (SRV) records from the at least one secondary server, wherein - if the second server for a service use by the client device is available - the network service provided to the client device by the second server becomes.

Description

STATE OF THE ART

The present invention relates to a method for improved load sharing in providing a network service in a computer network, the computer network comprising a plurality of servers for providing the network service to a client device, the plurality of servers comprising a plurality of primary servers, the plurality of servers including at least one secondary server Include failover.

Furthermore, the present invention relates to an improved load sharing system for providing a network service in a computer network, the system comprising the computer network and a client device, the computer network comprising a plurality of servers for providing the network service to a client device, the plurality of network servers Server comprises a plurality of primary servers, wherein the plurality of servers comprise at least one secondary server for failover.

Furthermore, the present invention relates to a computer program and a computer program product.

The Domain Name System (DNS) protocol provides a means of realizing load distribution in a network protocol (IP) based computer network in terms of network service provisioning and failover. The Domain Name System (DNS) provides Service Resource Record (SRV) records for load balancing and failover. The Service Resource Record (SRV) records contain a plurality of entries, each entry to a (requested) network service, including a number of other data fields (for example, the network address or host name of the server providing the requested network service), a prioritization value (which is also referred to as "priority"). and a load distribution value (also referred to as "weight"). The Service Resource Record (SRV) records can use this information in a Domain Name System (DNS) to provide load balancing and failover (by implementing redundant servers).

A disadvantage of load distribution methods implemented in such a way is the possible occurrence of a congestion situation of a server, if it is the penultimate available server of a group of servers providing a specific network service with the same prioritization value - i. E. if, for example, all the other servers in this group have failed, causing a deterioration in the quality of service of the network service and / or the congested server also fails.

DISCLOSURE OF THE INVENTION

It is an object of the present invention to provide a method for improved load sharing with respect to the provision of a network service in a computer network, by degrading the quality of service and / or the failure of a server of a group of servers having the same prioritization value - especially at the same time a relatively low configuration effort - is avoided.

• – Ermitteln, durch das Client-Gerät, eines ersten Servers, wobei der erste Server in Abhängigkeit der Lastverteilungsdaten der Service Resource Record (SRV) Datensätze aus den mehreren Primärservern selektiert wird,
• – Bereitstellen – für den Fall, dass der erste Server für die Bereitstellung des Netzwerkdienstes an das Client-Gerät verfügbar ist – des Netzwerkdienstes an das Client-Gerät durch den ersten Server, sowie – für den Fall, dass der erste Server für die Bereitstellung des Netzwerkdienstes an das Client-Gerät ausfällt und ein weiterer Server der mehreren Primärserver für die Bereitstellung des Netzwerkdienstes an das Client-Gerät verfügbar ist – Ermitteln, durch das Client-Gerät, eines zweiten Servers aus dem wenigstens einen Sekundärserver, wobei der zweite Server in Abhängigkeit der Lastverteilungsdaten der Service Resource Record (SRV) Datensätze aus dem wenigstens einen Sekundärserver ermittelt wird, wobei – falls der zweite Server für eine Dienstnutzung durch das Client-Gerät verfügbar ist – der Netzwerkdienst dem Client-Gerät durch den zweiten Server bereitgestellt wird.

The object of the present invention is achieved by a method for improved load distribution with respect to providing a network service in a computer network, wherein the computer network comprises a plurality of servers for providing the network service to a client device, wherein the plurality of servers comprises a plurality of primary servers, wherein the plurality of servers at least one failover secondary server, the computer network having a Domain Name System (DNS) server for providing Service Resource Record (SRV) records, the Service Resource Record (SRV) records having load distribution data and prioritization data, the client device having connected to the computer network for using the network service, the method comprising the steps of:

• Determining, by the client device, a first server, wherein the first server is selected from the plurality of primary servers in dependence on the load distribution data of the service resource record (SRV),
• Providing, in the case where the first server is available to provide the network service to the client device, the network service to the client device through the first server; and in the event that the first server for providing the network service Network service to the client device fails and another server of the multiple primary servers for the provision of the network service to the client device is available - determining, by the client device, a second server from the at least one secondary server, the second server depending the load distribution data of the Service Resource Record (SRV) records from the at least one secondary server is determined, wherein - if the second server for service use by the client device is available - the network service is provided to the client device by the second server.

As a result, according to the present invention, it is advantageously possible that a deterioration of the quality of service and / or further failure of a (further) server of a group of servers with the same prioritization value (in particular after a first server of this group of servers with the same prioritization value has already failed is) avoided by the second server is determined from the at least one secondary server. The multiple servers are configured to provide the network service to the client device and, in addition to the multiple primary servers, also includes the at least one secondary server for failover. Thus, in the event that the first server (a primary server) fails to deploy the network service to the client device and another primary server is available to provide the network service, the second server (ie a secondary server) becomes dependent on the load distribution data the Service Resource Record (SRV) determines records from the at least one secondary server. This means, in particular, that a failover server of the at least one secondary server (ie in particular the second server) is not available for selection as a replacement server (for the failed first server) only when all the primary servers have failed to provide the network service to the client device but is already available for selection as a replacement server for the failed primary server if at least one other primary server of the multiple primary servers is still available to provide the network service to the client device.

According to the present invention, it is provided in particular that the remaining servers of the multiple servers - from which a replacement server (for the failed first server) is determined as a function of failure prioritization data of the Service Resource Record (SRV) records - the plurality of primary servers - at least with the exception of the failed for the provision of the network service to the client device first server - and the at least one secondary server and in particular further failover servers.

In particular, in accordance with the present invention, provision of a server by the client device includes accessing the Service Resource Record (SRV) records (stored in the client device).

In particular, according to the present invention, it is contemplated that the one client device is also at least one client device at the same time - i. exactly one client device or multiple client devices. In particular, in the case of a plurality of client devices, the method steps according to the method of the present invention provide the logical sequence of the method steps for exactly one client device of the plurality of client devices, wherein the temporal sequence of the method steps different client Devices can also be different. In particular, a server is determined to use the network service provided by the determined server by the client device.

In particular, in accordance with the present invention, "in the event that a server fails to provide the network service to the at least one client device" means that the server is unavailable to provide the network service to the at least one client device is - ie For example, an existing server will fail or a destination server (according to the data field "target") in the Service Resource Record (SRV) records (which are stored in the client device) will not or no longer exist or will be labeled incorrectly.

• – Ermitteln, durch das Client-Gerät, eines weiteren ersten Servers, wobei der weitere erste Server in Abhängigkeit der (auf die Primärserver bezogenen) Lastverteilungsdaten der Service Resource Record (SRV) Datensätze aus den mehreren Primärservern selektiert wird,
• – Bereitstellen – für den Fall, dass der weitere erste Server für die Bereitstellung des Netzwerkdienstes an das Client-Gerät verfügbar ist – des Netzwerkdienstes an das Client-Gerät durch den weiteren ersten Server, sowie – für den Fall, dass der weitere erste Server für die Bereitstellung des Netzwerkdienstes an das Client-Gerät ausfällt – Ermitteln, durch das Client-Gerät, eines weiteren zweiten Servers, wobei der weitere zweite Server in Abhängigkeit der Priorisierungsdaten der Service Resource Record (SRV) Datensätze aus den übrigen Servern der mehreren Server – d.h. insbesondere aus den mehreren Servern mit Ausnahme des weiteren ersten Servers – ermittelt wird, wobei – falls der weitere zweite Server für eine Dienstnutzung durch das Client-Gerät verfügbar ist – der Netzwerkdienst dem Client-Gerät durch den weiteren zweiten Server bereitgestellt wird.

In particular, according to the present invention, in the event that the first server is available to provide the network service to the client device, the method further comprises the steps of:

• Determining, by the client device, a further first server, wherein the further first server is selected as a function of the (on the primary server related) load distribution data of the Service Resource Record (SRV) records from the plurality of primary servers,
• Providing, in the event that the further first server is available for the provision of the network service to the client device, the network service to the client device through the further first server, and - in the event that the further first server for the provisioning of the network service to the client device fails - determining, by the client device, another second server, wherein the further second server depending on the prioritization data of the service resource record (SRV) records from the remaining servers of the multiple servers - ie in particular from the multiple servers except the further first server - is determined, wherein - if the other second server for a service use by the client device is available - the network service is provided to the client device by the further second server.

According to the present invention, it is provided in particular that the second server or the second server - in the event that the first or the other first server for the provision of the network service to the client device fails - depending on the first server or other first server related failure prioritization value, depending on the primary server related load distribution data and depending on the at least one secondary server related load distribution data and prioritization data of the Service Resource Record (SRV) records is determined so that in particular the related to the first server or other first server default priority value with the second server related prioritization value is equal ,

According to a preferred embodiment of the present invention, it is provided that the at least one secondary server is determined from the plurality of servers as a function of a service prioritization value of the service resource record (SRV) records.

As a result, according to the present invention, it is possible in an advantageous manner for the service resource record (SRV) data sets from the at least one secondary server to be determined, in particular for determining the second server, the at least one secondary server from the plurality of servers as a function of a default priority value the Service Resource Record (SRV) records is determined. As a result, in particular, a secondary server is selected as a priority (i.e., prior to the primary servers, in particular even if at least one other primary server is still available) and only a junior server is selected (that is, only when all secondary servers fail to provide the network service).

According to a further preferred embodiment of the present invention, it is provided that - in the event that the first server fails to provide the network service to the client device and another server of the plurality of primary servers for the provision of the network service to the client device is available - only a server of at least one secondary server for failover for the first server is selected.

As a result, according to the present invention, it is advantageously possible to ensure that the second server (or spare server for the failed primary server) is a failover server, such that it is compromised by the first (primary) failed to provide the network service to the client device -) server does not come to an increased load of the other available primary server. Thus, an improved load distribution over the prior art is advantageously achieved in that as the second server, a replacement server from the group of (secondary) failsafe server is determined.

According to a preferred embodiment of the present invention, it is provided that the at least one secondary server comprises a first secondary server and a second secondary server, wherein - in the event that the first server for the provision of the network service to the client device fails and another server the multiple primary server is available for providing the network service to the client device - the first secondary server is determined as the second server from the at least one secondary server in response to the load distribution data of the Service Resource Record (SRV) records related to the secondary servers.

As a result, according to the present invention, it is advantageously possible for the requests for the network service initially distributed to the failed first server to be divided among several secondary servers, so that the second server has to process a lower network load compared with the failed first server. This server resources can be saved and still a comparatively high availability can be realized.

According to a preferred embodiment of the present invention, it is provided that the first server is determined from the plurality of primary servers in dependence on the load distribution data of the Service Resource Record (SRV) data records relating to the plurality of primary servers, wherein - in the event that the first server for the provisioning of the network service to the client device fails and another server of the plurality of primary servers is available to provide the network service to the client device and the second server is available for service use by the client device - the network service to the client device is provided by the second server such that a portion of network service requests of the client device is processed by the second server, the proportion being determined in dependence on a load distribution value related to the first server, the proportion depending in particular on a multiplication n of the load distribution value related to the first server is determined with a load distribution value related to the second server.

Thus, according to the present invention, it is advantageously possible that even if only the first server fails - but in particular still at least one other primary server is available - only the originally addressed to the failed primary server requests to the secondary server or fail-safe server according to the Redistribute load distribution data associated with the secondary servers. This advantageously loads all servers with less than 100% of the requests as long as at least two servers are available to provide the network service to the client device.

According to a preferred embodiment of the present invention, it is provided that the Service Resource Record (SRV) records comprise prioritization data, load distribution data, and outage prioritization data, wherein one entry of each of the Service Resource Record (SRV) records is associated with a server of the plurality of servers, each entry wherein the first server is determined as a function of a first prioritization value and a first load distribution value of a first entry associated with the first server, and - in the event that the first server for the provisioning of the network service to the client device fails and another server of the multiple primary servers is available to provide the network service to the client device - the second one Server is determined in response to a first failure prioritization value of the first entry.

As a result, according to the present invention, it is advantageously possible to minimize the configuration effort required to avoid deterioration of the quality of service and / or the failure of a server of the group of servers with the same prioritization value.

According to a preferred embodiment of the present invention, it is provided that - in the event that the first server fails to provide the network service to the client device and another server of the multiple primary servers for the provision of the network service to the client device available in that a server is determined as the second server to which a second prioritization value assigned in a second entry of the Service Resource Record (SRV) records has the same priority as a first service prioritization value of the Service Resource Record (SRV) records assigned to the first server.

As a result, according to the present invention, it is advantageously possible to minimize the configuration effort required to avoid deterioration of the quality of service and / or failure of a server of the group of servers with the same prioritization value, according to the invention improved load distribution is achieved in particular with relatively little configuration effort by the Domain Name System (DNS) protocol is extended by an additional data field per entry. In particular, the additional data field is referred to as the outage prioritization value and the entirety of the outage prioritization values of all entries of the Service Resource Record (SRV) data records are also referred to as outage prioritization data.

According to a preferred embodiment of the present invention, it is provided that - in the event that the first server for the provision of the network service to the client device fails and all secondary servers - which in the service resource record records, in particular, a prioritization value is assigned is equal to the first failure prioritization value of the first server - failing to provide the network service to the client device - failover for the failed first server another first server is determined from the multiple primary servers, whose prioritization value in particular with the first prioritization value of the first server is equal.

As a result, according to the present invention, it is advantageously possible for secondary servers (i.e., servers having a prioritization value equal to the failed first server default priority value) to be a substitute server for the failed first server, and firstly subordinate - i.e. in the event that all secondary servers have failed to deploy the network service - another primary server still available is used. As a result, an improved load distribution over the prior art is achieved.

According to a preferred embodiment of the present invention, it is provided that - in the event that the first server for the provision of the network service to the client device fails and all secondary servers - especially with an equal to the first default prioritization value of the first server prioritization value - for the provisioning of the network service to the client device fails and all further primary servers - in particular with a prioritization value equal to the first prioritization value of the first server - fail to provide the network service to the client device - failover for the failed first server a tertiary server of a plurality of servers is determined whose priority value is subordinate to the first default priority value of the first server in particular

As a result, according to the present invention, it is advantageously possible that, in addition to improved load distribution, an improved failover can also be achieved by using the tertiary servers only when all primary servers and secondary servers have failed.

Furthermore, the present invention relates to a system for improved load distribution relating to the provision of a network service in a computer network, the system comprising the computer network and a client device, the computer network comprising a plurality of servers for providing the network service to the client device, the plurality of servers comprising a plurality of primary servers, the plurality of servers at least a secondary server for failover, the computer network having a Domain Name System (DNS) server for providing Service Resource Record (SRV) records, the Service Resource Record (SRV) records having load distribution data and prioritization data, the client device having the Computer network is connected to use the network service, wherein the client device is configured to determine a first server, wherein the client device is configured, the first server depending on the load distribution data of the Service Resource Record (SRV) records from the plurality To select primary servers, wherein the system is configured to provide the network service to the client device through the first server in case the first server is available to provide the network service to the client device; in the event that the first server fails to provide the network service to the client device and another server of the multiple primary servers is available to provide the network service to the client device - by the client device a second server from the at least one a secondary server is determined, wherein the client device is configured such that the second server is determined in dependence on the load distribution data of the Service Resource Record (SRV) records from the at least one secondary server, the system being configured such that - in the case in that the second server is for service use by the client device t is available - the network service is provided to the client device by the second server.

Thereby, according to the present invention, it is advantageously possible to provide a system with a computer network and a client device configured such that a deterioration of the quality of service and / or a failure of a server of a group of servers with the same prioritization value thereby avoided is that the second server is determined from the at least one secondary server. The multiple servers are configured to provide the network service to the client device and, in addition to the multiple primary servers, also includes the at least one secondary server for failover. Thus, in the event that the first server fails to provide the network service to the client device and another server of the multiple primary servers is available to provide the network service to the client device, the second server becomes the second server depending on the load distribution data Service Resource Record (SRV) records from the at least one secondary server determined. In particular, this means that the system is configured such that a failover server of the at least one secondary server (that is, in particular the secondary server) is not available for selection as a replacement server until all primary servers for providing the network service to the client device have failed but is already available for selection as a replacement server for the failed primary server if at least one other primary server of the multiple primary servers for the provision of the network service to the client device are available.

According to a preferred embodiment of the present invention, it is provided that the system is configured such that, in the event that the first server for providing the network service to the client device fails and another server of the plurality of primary servers for the provision of Network service to the client device is available - only one server of at least one secondary server for failover for the first server is selected.

As a result, according to the present invention, it is advantageously possible for the system according to the invention to ensure that the second server (or spare server for the failed primary server) is a fail-safe server, thus causing a failure of the server to provide the network service to the client device originally intended primary (primary) server does not result in increased load on the other available primary server. Thus, an improved load distribution in the system over the prior art is advantageously achieved in that the second server, a replacement server from the group of fail-safe servers is determined.

According to a preferred embodiment of the present invention, it is provided that the at least one secondary server comprises a first secondary server and a second secondary server, wherein the client device is configured such that - in the event that the first server for the provision of the network service the client device fails and another server of the multiple primary servers is available to provide the network service to the client device - the first secondary server as the second server from the at least one secondary server depending on the secondary server related load distribution data the service resource record ( SRV) records is determined.

Thereby, according to the present invention, it is advantageously possible to provide a system in which the client device's requests originally distributed to the failed first server are shared among multiple failover servers, so that the second server has less network load against the failed first server has to process. This server resources can be saved and still a comparatively high availability can be realized.

According to a preferred embodiment of the present invention, it is provided that the client device is configured such that the first server is determined from the plurality of primary servers in dependence on the load distribution data of the service resource record (SRV) data records relating to the plurality of primary servers Client device is configured such that - in the event that the first server fails to provide the network service to the client device and another server of the multiple primary server for the provision of the network service to the client device is available and the second Server is available for service use by the client device - the network service is provided to the client device by the second server such that a share of network service requests of the client device is processed by the second server, the proportion depending on the first server related Lastv the share is determined in particular as a function of a multiplication of the load distribution value related to the first server with a load distribution value related to the second server.

As a result, according to the present invention, it is advantageously possible for the system to be configured such that even if only the first server fails, but in particular still at least one other primary server is available, only the requests originally addressed to the failed primary server the secondary servers or failover servers are redistributed according to the load distribution data associated with the secondary servers. This advantageously loads all servers with less than 100% of the requests as long as at least two servers are available to provide the network service to the client device.

Furthermore, the present invention relates to a computer program with computer-readable program code, by means of which the inventive method is feasible when the computer-readable program code is executed on a programmable device.

Furthermore, the present invention relates to a computer program product with a computer program stored on a storage medium according to the invention, with the aid of which a method according to the invention can be carried out if the computer-readable program code is executed on a programmable device.

These and other features, features, and advantages of the present invention will become apparent from the following detailed description when taken in conjunction with the accompanying drawings which illustrate, by way of example, the principles of the invention. The description is by way of example only, without limiting the scope of the invention. The references cited below refer to the attached drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

1 schematically illustrates a system for load distribution with respect to the provision of a network service in a computer network according to the present invention.

2 schematically illustrates a system for load distribution relating to the provision of a network service in a computer network.

3 schematically illustrates a system for improved load distribution with respect to the provision of a network service in a computer network according to the present invention.

4 schematically illustrates a system for improved load distribution with respect to the provision of a network service in a computer network according to the present invention.

5 schematically illustrates a system for improved load distribution with respect to the provision of a network service in a computer network according to the present invention.

EMBODIMENT (S) OF THE INVENTION

In 1 is a load balancing system for providing a network service in a computer network 100 illustrated schematically according to the present invention. The system includes the computer network 100 and a client device 40 , The client device 40 is with the computer network 100 connected to the computer network 100 provided network service by the client device 40 is being used.

The computer network 100 includes several servers 11 . 12 . 21 . 22 to provide the network service to the client device 40 where the multiple servers 11 . 12 . 21 . 22 both multiple primary servers 11 . 12 , as well as at least one secondary server 21 . 22 for failover. In 1 are two primary servers 11 . 12 the multiple primary server and two secondary servers 21 . 22 the at least one secondary server 21 . 22 shown.

Furthermore, the computer network includes 100 a Domain Name System (DNS) server 50 to provide Service Resource Record (SRV) records. In particular, the Service Resource Record (SRV) records contain a plurality of entries, each entry (or record) to the network service including a number of other data fields (including, for example, the network address or host name of the server providing the network service) a prioritization value (which also is referred to as "priority") and a load distribution value (which is also referred to as "weight"). Here, the totality of the prioritization values of the multiple entries of the Service Resource Record (SRV) data sets is also referred to as prioritization data, and correspondingly the totality of the load distribution values is also referred to as load distribution data.

According to a preferred embodiment of the present invention, the service resource record (SRV) records comprise a plurality of entries, wherein an entry of the service resource record (SRV) records has in particular the following structure:
"_Service._proto.name. TTL class SRV priority weight port target. ".

• – service: Der symbolische Name des (angefragten) Netzwerkdienstes,
• – proto: Das Transportprotokoll des (angefragten) Netzwerkdienstes, wobei beispielsweise das Transportprotokoll entweder das Transmission Control Protocol (TCP) oder das User Datagram Protocol (UDP) Netzwerkprotokoll ist,
• – name: Der Domainname für den der Eintrag gültig ist,
• – TTL: Die Gültigkeitsdauer (die insbesondere vorgibt, wie lange der Eintrag in einem Domain Name System (DNS) Cache des Client-Geräts 40 gehalten werden darf),
• – class: Das Klassenfeld, wobei insbesondere „IN“ für Internet steht,
• – priority: Der Priorisierungswert (bzw. „Priorität“), welcher einem im Datenfeld „target“ bezeichneten Server (oder Host) – welcher auch als Zielserver bezeichnet wird – zugeordnet ist,
• – weight: Der Lastverteilungswert (bzw. „Gewicht“), der gemäß der vorliegenden Erfindung insbesondere ein relatives (Wahrscheinlichkeits-)Gewicht für Einträge mit derselben Priorität spezifiziert, wobei insbesondere Lastverteilungswerte mit höherem Zahlenwert insbesondere eine proportional höhere Wahrscheinlichkeit gegeben werden soll, durch das Client-Gerät 40 ausgewählt zu werden,
• – port: Der TCP oder UDP Port (des Zielservers), an dem der Netzwerkdienst bereitgestellt wird,
• – target: Der Server (Zielserver), welcher den Netzwerkdienst (welcher im Datenfeld „sevice“ bezeichnet ist) im Computernetzwerk 100 bereitstellt bzw. anbietet, wobei es sich bei dem Datenfeld „target“ insbesondere um den kanonischen Hostname des Geräts handelt, welches den Netzwerkdienst im Computernetzwerk 100 bereitstellt.

According to a preferred embodiment of the present invention, each entry of the Service Resource Record (SRV) data records comprises at least the following data fields:

• - service: The symbolic name of the (requested) network service,
• Proto: The transport protocol of the (requested) network service, where, for example, the transport protocol is either the Transmission Control Protocol (TCP) or the User Datagram Protocol (UDP) network protocol,
• - name: the domain name for which the entry is valid,
• - TTL: The period of validity (which in particular dictates how long the entry in a Domain Name System (DNS) cache of the client device 40 may be held),
• - class: the class field, whereby in particular "IN" stands for Internet,
• Priority: The prioritization value (or "priority") associated with a server (or host) designated in the data field "target" - which is also referred to as the destination server.
• Weight: The load distribution value (or "weight") which, according to the present invention, specifies in particular a relative (probability) weight for entries having the same priority, in particular load distribution values with a higher numerical value, in particular a proportionally higher probability, being given by client device 40 to be selected
• - port: The TCP or UDP port (of the destination server) where the network service is deployed,
• Target: The server (destination server) that hosts the network service (which is denoted in the sevice data field) on the computer network 100 In particular, the data field "target" is the canonical host name of the device that hosts the network service on the computer network 100 provides.

According to the present invention, the client device is 40 in particular, configured such that the client device 40 - to a given network service (for example, a Session Initiation Protocol (SIP) service) - must try to contact the server with the lowest prioritization value that the client device 40 in the computer network 100 in particular, the client device 40 is configured such that a plurality of destination servers are contacted with the same prioritization value (or "priority") in an order and / or according to a frequency distribution, which is determined as a function of the load distribution value (or the weight or "weight") Load distribution value indicates the likelihood that the client device 40 (next) selects the target server (associated with the entry) from the available servers with the same prioritization value).

In 1 an exemplary embodiment is shown in which the client device 40 at the Domain Name System (DNS) server 50 retrieved Service Resource Record (SRV) records of a zone file containing the following four entries:
_sip._tcp.example.com. 3600 IN SRV 0 20 5060 server11.example.com
_sip._tcp.example.com. 3600 IN SRV 0 80 5060 server12.example.com
_sip._tcp.example.com. 3600 IN SRV 1 50 5060 server21.example.com
_sip._tcp.example.com. 3600 IN SRV 1 50 5060 server22.example.com

The Service Resource Record (SRV) records the client device 40 the information transmitted that for the requested network service - here the Session Initiation Protocol (SIP) service - primarily or primarily the server 11 and 12 with the lowest priority value (which is here "0"). Therefore, the servers 11 and 12 with the lowest priority value here as the primary server 11 . 12 designated. The primary servers 11 . 12 here are TCP servers, whereby in each case the haven 5060 is to be used. According to the present invention, the network service requests (here the SIP requests) of the client device become 40 on the two primary servers 11 . 12 in proportion to the load distribution values distributed in the Service Resource Record (SRV) records to the primary servers 11 . 12 associated with the lowest priority value. Here are the network service requests in the ratio 20:80 on the two primary servers 11 . 12 distributed. Thus, a load distribution based on the Domain Name System (DNS) protocol is realized. This load distribution situation is in 1 by the with reference number 1 illustrated double arrow illustrated. In the event that all servers with the same prioritization value (in this example, the two primary servers 11 . 12 with the prioritization value "0") for providing the network service to the client device 40 fail, those become servers 21 . 22 through the client device 40 which has a subordinate prioritization value - here a higher prioritization value "1". These are in the in 1 shown exemplary embodiment, the two secondary servers 21 and 22 , The one with reference number 3 indicated arrow illustrates the failure of all primary servers 11 . 12 - not just a single primary server. In the process, the network service requests (here the SIP requests) become the secondary servers 21 . 22 (here the two TCP servers 21 and 22 ) is distributed in a ratio corresponding to the one in the Service Resource Record (SRV) records each to the secondary servers 21 . 22 corresponding load distribution values corresponds - here in the ratio 50:50. This failover is in 1 by the with reference number 2 illustrated double arrow illustrated.

In 2 Fig. 12 schematically illustrates a load sharing system for providing a network service in a computer network. Through the client device 40 becomes first a first server 11 depending on the load distribution data of the Service Resource Record (SRV) records from the multiple primary servers 11 . 12 - ie in particular from the servers 11 . 12 with the lowest prioritization value - determined. In the event that the first server 11 to provide the network service to the client device 40 is available, the network service is through the first server 11 to the client device 40 provided (this case is in 2 not illustrated). However, if the case occurs (see 2 ) that the first server 11 to provide the network service to the client device 40 fails, is through the client device 40 , a replacement server to replace the failed first server 11 (for example, the second server 21 ). This means in particular that the first server 11 to provide the network service to the client device 40 that fails the first server 11 (at least for the client device 40 ) is unavailable - for example due to a server failure or due to changes in the structure of the computer network 100 or other circumstances that are not caused by the client device 40 existing Service Resource Record (SRV) records. The replacement server is determined here as a function of the prioritization data of the Service Resource Record (SRV) data records.

Especially if in the case described last only a single primary server 12 - which has the same prioritization value as the failed first server 11 owns - is available and only the remaining primary server 12 if records were selected as the only server according to the entries of the Service Resource Record (SRV), then (in this relatively unfavorable case) the selected primary server would 12 as the only server to provide the network service to the client device 40 be used and thus a network load of 100% of all requests by the client device 40 need to process the secondary server 21 . 22 no network service requests would be received. Only in the event of a failure of the hitherto remaining primary server 12 In this case, it would handle the requests of the client device 40 through the secondary server 21 . 22 come.

In contrast, according to the present invention, the spare server becomes the other servers 12 . 21 . 22 the multiple server 11 . 12 . 21 . 22 determined, which in particular include both primary server, and failover server. Here are the remaining servers 12 . 21 . 22 the remaining primary servers of the multiple primary servers 12 Ie all available primary servers 12 with the exception of the failed first server 11 - As well as the at least one secondary server 21 . 22 and especially other failover servers. This makes it possible according to the invention in an advantageous manner that the described congestion situation of the last remaining server (here the primary server 12 ) is avoided, leaving the primary server 12 here not the entire network load on the requests from the client device 40 must process. According to the present invention, this advantageously achieves a load distribution in such a way that all available servers - in particular servers with the same prioritization value and servers with different prioritization values (including the failover servers) - are always used for load distribution, even if all servers have the same prioritization value (here primary server 11 . 12 ) at least one server fails and at least one server is available at the same time.

In 3 For example, a system for improved load sharing in providing a network service in a computer network in accordance with the present invention is illustrated schematically. The inventively improved load distribution is realized for a system which in particular already uses the Domain Name System (DNS) protocol for load distribution and failover. According to a preferred embodiment of the present invention, the Domain Name System (DNS) protocol is extended by an additional data field per entry. In particular, the additional data field is referred to as the outage prioritization value and the entirety of the outage prioritization values of all entries of the Service Resource Record (SRV) data records are also referred to as outage prioritization data.

According to a preferred embodiment of the present invention, the entries of the Service Resource Record (SRV) data records each essentially have the following form:
"_Service._proto.name. TTL class SRV priority weight failure_priority port target. ".

As a position of the additional data field - which is referred to here as "failure_priority" - may alternatively be selected in particular another position within the entry.

According to a preferred embodiment of the present invention, the failure prioritization value (or the data field "failure_priority"), in the event that the destination server according to the data field "target" (for the provision of the network service according to the data field "_service"), is sent to the client computer. device 40 ) - defines the prioritization value (or data field "priority") of an alternative server (second server or spare server). This means, according to a preferred embodiment of the present invention, a configuration of the client device 40 such that - in the event that the first server 11 to provide the network service to the client device 40 fails - as a replacement server for the failed first server 11 a server is determined as a function of the default prioritization data of the service resource record records such that the in the service resource record records the (failed) first server 11 assigned default priority value and the prioritization value assigned to the spare server in the service resource record records are equal. This situation is exemplary in 3 by the with reference number 4 illustrated double arrow illustrated.

According to an exemplary embodiment according to 3 For example, the Service Resource Record (SRV) records of the present invention include the following four entries:
_sip._tcp.example.com. 3600 IN SRV 0 80 1 5060 server11.example.com
_sip._tcp.example.com. 3600 IN SRV 0 20 1 5060 server12.example.com
_sip._tcp.example.com. 3600 IN SRV 1 50 1 5060 server21.example.com
_sip._tcp.example.com. 3600 IN SRV 1 50 1 5060 server22.example.com.

Here is - in the event that as the first server one of the two primary servers 11 or 12 with the lowest prioritization value "0" (in 3 this is the primary server 11 ) - as a second server, a destination server from the secondary servers 21 . 22 is determined. In this exemplary embodiment, in particular, only one secondary server is used 21 . 22 because the first entry has the default priority value "1" and the two secondary servers 21 . 22 have the prioritization value "1" corresponding to the default failure value of the failed first server 11 is equal. In the event that the determined second server 21 for service usage by the client device 40 is available, the network service becomes the client device 40 provided by the second server. This situation is by reference number 2 ' in 3 illustrated.

According to a preferred embodiment of the present invention - in the event that the first server 11 to provide the network service to the client device 40 fails - the network service requests by the client device 40 distributed according to the load distribution data on the other servers. The network service requests are distributed according to the load distribution data related to the primary servers. In the example given above, these are 20% of the requests made to the primary server according to the weight "20" 12 be distributed. The remaining requests - that is, the requests according to the load distribution value of the first entry on the failed first server 11 would be omitted (in the above example, these are 80% of the requests by weight "80") - will be applied to the secondary servers according to the load distribution values 21 . 22 which are related to the secondary servers (in the example given above, according to the weights "50", these each account for 50% of the 80% of the requests (originally distributed to the failed first server), ie in total 40% of the requests ). Especially in the event that all primary servers 11 . 12 fail, get here the two secondary servers 21 . 22 50% of the requests from the client device 40 , As a result, according to the invention, it is advantageously possible for all servers to process less than 100% of the network load as long as at least two servers are available.

In 4 is a system for improved load sharing with respect to the provision of a Network service in a computer network according to the present invention schematically illustrates. According to a preferred embodiment of the present invention is - in the event that all the target server with a failed to the first server 11 Equal prioritization value (here primary server 11 . 12 ) for providing the network service to the client device 40 fail (or are unavailable) and beyond all target server (here secondary server 21 . 22 ) for providing the network service to the client device 40 fail, to which a prioritization value is assigned in the Service Resource Record (SRV) records, with the prioritization value of the (failed) first server 11 is equal to - a server is determined as the replacement server, which is assigned a prioritization value in the Service Resource Record (SRV) records, compared to the default priority value of the (failed) first server 11 subordinate (here larger). In the in 4 Exemplified embodiments are all primary servers 11 . 12 failed and beyond all secondary servers 21 . 22 so that another (in 4 however, not shown) server with a prioritization value (in this example, this would be the value "2" or more), compared to the first server in the Service Resource Record records 11 assigned default priority value (in this example the value "1") is subordinate. In this example, such another server would be a tertiary server used as a spare server.

According to a preferred embodiment of the present invention, the computer network comprises 100 multiple primary servers 11 . 12 with a first primary server 11 and a second primary server 12 , Furthermore, the computer network comprises at least one secondary server 21 . 22 (ie failover server) with a first secondary server 21 and a second secondary server 22 , and in particular other failover servers. Through the client device 40 are used by the Domain Name System (DNS) server 50 (specific for use by the client device 40 selected) Service Resource Record (SRV) records are retrieved and stored in a Domain Name System (DNS) cache of the client device 40 saved.

According to an exemplary embodiment of the present invention, the Service Resource Record (SRV) records - for the same network service - include the first primary server 11 associated first entry, a second primary server 12 associated second entry, a first secondary server 21 associated third entry and a second secondary server 22 associated fourth entry. The first entry - which is the first primary server 11 is assigned - has a first prioritization value, a first load distribution value, and a first default prioritization value. The second entry - which is the second primary server 12 - has a second prioritization value, a second load distribution value, and a second default prioritization value. The third entry - which is the first secondary server 21 - has a third prioritization value, a third load distribution value, and a third default prioritization value. The fourth entry - which is the second secondary server 22 - has a fourth prioritization value, a fourth load distribution value and a fourth default prioritization value. In particular, the Service Resource Record (SRV) records include a fifth entry - which is associated with the tertiary server - and a fifth prioritization value, a fifth load distribution value, and a fifth default prioritization value.

According to an exemplary embodiment of the present invention, the first and second prioritization values have the same value (for example, the value "0"). Furthermore, the third and the fourth prioritization value have the same value (for example the value "1"), which is subordinate to the first and second prioritization value (in this case, higher here). Further, the fifth prioritization value is subordinate to the first, second, third, and fourth prioritization values (i.e., the fifth prioritization value is "2", for example). The first load distribution value in this exemplary embodiment is 80, the second load distribution value is 20, the third load distribution value is 50, the fourth load distribution value is 50, and more specifically, the fifth load distribution value is 100. Further, here, the first to fourth failure prioritization values each have a value is equal to the third and fourth prioritization values - in this exemplary embodiment this is the value "1".

According to an exemplary embodiment of the present invention - in the event that the first and second primary servers 11 . 12 available - a first portion of the requests for the network service according to the first load distribution value (here 80% of the requests) by the first primary server 11 and a second portion of the network service requests according to the second load distribution value (here 20% of the requests) by the second primary server 12 processed. In the event that the first primary server 11 to provide the network service to the client device 40 fails and the second primary server 12 is still available, according to the present invention, the second portion of requests for the network service according to the second load distribution value (here 20% of the requests) will continue to be made by the second primary server 12 Since the first failure prioritization value (here "1") and the second and third prioritization value (here also "1") have the same value, the first portion of the network service requests are calculated according to the first load distribution value (here 80% of the requests). for processing by the first and second secondary servers 21 . 22 on the first and second secondary servers 21 . 22 distributed. Thus - in the event that the first server 11 to provide the network service to the client device 40 fails - the second server 21 determined as a function of the first failure prioritization value (here "1") of the first entry. The first and second secondary servers 21 . 22 Inquiries (here totaling 80%) are made in the ratio of the third and fourth load distribution values (ie in each case 50%) to the first and second secondary servers 21 . 22 distributed. Thus - in the event that the first server 11 to provide the network service to the client device 40 fails - the second server 21 from one of the two secondary servers 21 . 22 depending on the (the secondary servers assigned) load distribution data of the Service Resource Record (SRV) records selected.

In 5 is a preferred embodiment of the present invention illustrated, wherein - in the event that the first primary server 11 to provide the network service to the client device 40 fails, in addition, the first and second secondary server 21 . 22 to provide the network service to the client device 40 fail and also all other secondary servers for the provision of the network service to the client device 40 fail (ie, such other secondary servers to which a prioritization value is assigned in the Service Resource Record (SRV) records that corresponds to the first failure prioritization value of the first server 11 is the same) - as replacement server, the second primary server 12 determines its prioritization value with the first prioritization value of the first server 11 is equal. It will only then (ie subordinate) if all (priority) secondary server 21 . 22 to provide the network service to the client device 40 have failed, a primary server 12 selected.

• – der erste Primärserver 11 für die Bereitstellung des Netzwerkdienstes an das Client-Gerät 40 ausfällt, und
• – alle übrigen Server 21, 22 mit einem zu dem ersten Ausfallpriorisierungswert des ersten Servers 11 gleichrangigen (dritten bzw. vierten) Priorisierungswert für die Bereitstellung des Netzwerkdienstes an das Client-Gerät 40 ausfallen (hier der erste und zweite Sekundärserver 21, 22), und darüber hinaus
• – alle übrigen Server 12 mit einem zu dem ersten Priorisierungswert des ersten Servers 11 gleichrangigen Priorisierungswert für die Bereitstellung des Netzwerkdienstes an das Client-Gerät 40 ausfallen (hier der zweite Primärserver 12), durch das Client-Gerät als Ersatzserver aus den mehreren Server 11, 12, 21, 22 ein Server selektiert, dessen Priorisierungswert gegenüber dem ersten Ausfallpriorisierungswert des ersten Servers 11 nachrangig ist. Das bedeutet hier, dass der Tertiärserver selektiert wird, bei welchem der fünfte Priorisierungswert (hier der Wert „2“) gegenüber dem ersten Ausfallpriorisierungswert (mit dem Wert „1“) nachrangig ist (d.h. hier ist der fünfte Priorisierungswert „2“ größer als der erste Ausfallpriorisierungswert „1“).

According to a preferred embodiment of the present invention (which in 4 is schematically illustrated) is used in the event that

• - the first primary server 11 to provide the network service to the client device 40 fails, and
• - all other servers 21 . 22 with a to the first failure prioritization value of the first server 11 Equal (third or fourth) prioritization value for the delivery of the network service to the client device 40 fail (here the first and second secondary server 21 . 22 ), and beyond
• - all other servers 12 with a to the first prioritization value of the first server 11 Equal prioritization value for the delivery of the network service to the client device 40 fail (here the second primary server 12 ), by the client device as a spare server from the multiple servers 11 . 12 . 21 . 22 a server selects its prioritization value against the first outage priority value of the first server 11 is subordinate. This means that the tertiary server is selected in which the fifth prioritization value (here the value "2") is subordinate to the first default prioritization value (with the value "1") (ie, the fifth prioritization value "2" is greater than that first failure prioritization value "1").

Claims (15)

Method for improved load distribution with regard to the provision of a network service in a computer network ( 100 ), whereby the computer network ( 100 ) several servers ( 11 . 12 . 21 . 22 ) for providing the network service to a client device ( 40 ), the multiple servers ( 11 . 12 . 21 . 22 ) multiple primary servers ( 11 . 12 ), wherein the multiple servers ( 11 . 12 . 21 . 22 ) at least one secondary server ( 21 . 22 ) for failover, whereby the computer network ( 100 ) a Domain Name System (DNS) server ( 50 ) for providing Service Resource Record (SRV) records, the Service Resource Record (SRV) records having load distribution data and prioritization data, the client device ( 40 ) with the computer network ( 100 ) for using the network service, the method comprising the steps of: - determining, by the client device ( 40 ), a first server ( 11 ), where the first server ( 11 ) depending on the load distribution data of the Service Resource Record (SRV) records from the multiple primary servers ( 11 . 12 ), - providing - in case the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) is available - the network service to the client device ( 40 ) through the first server ( 11 ), as well as - in case the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - determining, by the client device ( 40 ), a second server ( 21 ) from the at least one secondary server ( 21 . 22 ), the second server ( 21 ) depending on the load distribution data of the Service Resource Record (SRV) records from the at least one secondary server ( 21 . 22 ) is determined, where - if the second Server ( 21 ) for service use by the client device ( 40 ) - the network service is available to the client device ( 40 ) through the second server ( 21 ) provided. The method of claim 1, wherein the at least one secondary server ( 21 . 22 ) from the multiple servers ( 11 . 12 . 21 . 22 ) is determined as a function of an outage prioritization value of the Service Resource Record (SRV) records. Method according to one of the preceding claims, wherein - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) - only one server of the at least one secondary server ( 21 . 22 ) for failover for the first server ( 11 ) is selected. Method according to one of the preceding claims, wherein the at least one secondary server ( 21 . 22 ) a first secondary server ( 21 ) and a second secondary server ( 22 ), wherein - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - the first secondary server ( 21 ) as the second server ( 21 ) from the at least one secondary server ( 21 . 22 ) depending on the secondary server ( 21 . 22 ) determined load distribution data of the Service Resource Record (SRV) records is determined. Method according to one of the preceding claims, wherein the first server ( 11 ) from the multiple primary servers ( 11 . 12 ) depending on the multiple primary servers ( 11 . 12 determined load distribution data of the Service Resource Record (SRV) records, where - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available and the second server ( 21 ) for service use by the client device ( 40 ) - the network service is available to the client device ( 40 ) through the second server ( 21 ) such that a share of network service requests of the client device ( 40 ) through the second server ( 21 ), whereby the proportion depends on one on the first server ( 11 determined load distribution value, the proportion in particular as a function of a multiplication of the on the first server ( 11 ) load distribution value with one on the second server ( 21 ) determined load distribution value is determined. Method according to one of the preceding claims, wherein the Service Resource Record (SRV) records comprise prioritization data, load distribution data and default prioritization data, wherein in each case one entry of the Service Resource Record (SRV) records is sent to one server of the several servers (SRV). 11 . 12 . 21 . 22 Each entry includes a prioritization value of the prioritization data, a load distribution value of the load distribution data, and an outage prioritization value of the outage prioritization data, wherein the first server ( 11 ) in response to a first prioritization value and a first load distribution value of a first server ( 11 ) is determined and, in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - the second server ( 21 ) is determined as a function of a first default prioritization value of the first entry. Method according to one of the preceding claims, wherein - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - as a second server ( 21 ) a server is determined, to which in a second entry of the Service Resource Record (SRV) data records a second prioritization value is assigned, which is assigned to the first server ( 11 ) is ranked equal to the first outage prioritization value of the Service Resource Record (SRV) records. Method according to one of the preceding claims, wherein - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and all Secondary server ( 21 . 22 In particular, in the service resource record (SRV) data records, a prioritization value is associated with the first failure prioritization value of the first server. 11 ) is of equal importance - for the provision of the network service to the client device ( 40 ) fails - for failover for the failed first server ( 11 ) another first server ( 12 ) from the multiple primary servers ( 11 . 12 ) whose prioritization value is determined in particular with the first prioritization value of the first server ( 11 ) is of equal rank. Method according to one of the preceding claims, wherein - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and all secondary servers ( 21 . 22 ) - in particular with a first failure prioritization value of the first server ( 11 Equal Prioritization Value - to provide the network service to the client device ( 40 ) fails - for failover for the failed first server ( 11 ) a tertiary server of the several servers ( 11 . 12 . 21 . 22 ) whose prioritization value is determined in particular with respect to the first failure prioritization value of the first server ( 11 ) is subordinate. System for improved load distribution with regard to the provision of a network service in a computer network ( 100 ), the system hosting the computer network ( 100 ) and a client device ( 40 ), wherein the computer network ( 100 ) several servers ( 11 . 12 . 21 . 22 ) for providing the network service to the client device ( 40 ), the multiple servers ( 11 . 12 . 21 . 22 ) multiple primary servers ( 11 . 12 ), wherein the multiple servers ( 11 . 12 . 21 . 22 ) at least one secondary server ( 21 . 22 ) for failover, whereby the computer network ( 100 ) a Domain Name System (DNS) server ( 50 ) for providing Service Resource Record (SRV) records, the Service Resource Record (SRV) records having load distribution data and prioritization data, the client device ( 40 ) with the computer network ( 100 ) is connected to use the network service, the client device ( 40 ) for determining a first server ( 11 ), the client device ( 40 ), the first server ( 11 ) depending on the load distribution data of the Service Resource Record (SRV) records from the multiple primary servers ( 11 . 12 ), the system being configured such that, in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) - the network service to the client device ( 40 ) through the first server ( 11 ), and in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - through the client device ( 40 ) a second server ( 21 ) from the at least one secondary server ( 21 . 22 ), whereby the client device ( 40 ) is configured such that the second server ( 21 ) depending on the load distribution data of the Service Resource Record (SRV) records from the at least one secondary server ( 21 . 22 ), the system being configured such that, in the event that the second server ( 21 ) for service use by the client device ( 40 ) - the network service is available to the client device ( 40 ) through the second server ( 21 ) provided. The system of claim 10, wherein the system is configured such that, in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) - only one server of the at least one secondary server ( 21 . 22 ) for failover for the first server ( 11 ) is selected. The system according to one of claims 10 or 11, wherein the at least one secondary server ( 21 . 22 ) a first secondary server ( 21 ) and a second secondary server ( 22 ), wherein the client device ( 40 ) is configured such that - in the event that the first server ( 11 ) for providing the network service to the at least one client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available - the first secondary server ( 21 ) as the second server ( 21 ) from the at least one secondary server ( 21 . 22 ) depending on the secondary server ( 21 . 22 ) determined load distribution data of the Service Resource Record (SRV) records is determined. A system according to any one of claims 10 to 12, wherein the client device ( 40 ) is configured such that the first server ( 11 ) from the multiple primary servers ( 11 . 12 ) depending on the multiple primary servers ( 11 . 12 determined load distribution data of the Service Resource Record (SRV) records, the client device ( 40 ) is configured such that - in the event that the first server ( 11 ) for the delivery of the network service to the client device ( 40 ) fails and another server ( 12 ) of the multiple primary servers ( 11 . 12 ) for the delivery of the network service to the client device ( 40 ) is available and the second server ( 21 ) for service use by the client device ( 40 ) - the network service is available to the client device ( 40 ) through the second server ( 21 ) such that a share of network service requests of the client device ( 40 ) through the second server ( 21 ), whereby the proportion depends on one on the first server ( 11 determined load distribution value, the proportion in particular as a function of a multiplication of the on the first server ( 11 ) load distribution value with one on the second server ( 21 ) determined load distribution value is determined. Computer program with computer-readable program code, with the aid of which Method according to one of claims 1 to 9 is feasible when the computer readable program code is executed on a programmable device. A computer program product having a computer program stored on a storage medium according to claim 14, by means of which a method according to one of claims 1 to 9 can be carried out when the computer readable program code is executed on a programmable device.

Images