User plane function selection based on per subscriber CPU and memory footprint for packet inspection
Techniques are provided to enable the SMF or other network node performing UPF selection to determine a computational resource demand, also referred to herein as the computational footprint, associated with a PDU session. The computational footprint can then be used by the SMF or other network node to inform the selection of the UPF.
1. A method implemented by a network node in a wireless communication network to support selection of a user plane function (UPF) by a session management function (SMF) for a packet data session, the method comprising:
collecting one or more inspection parameters associated with packet inspection for a subscriber, a service, or both, for determining a per subscriber and/or or per service computational footprint associated with the packet inspection, wherein the computational footprint is indicative of the computational resources needed to perform packet inspection for one or more traffic flows included in the packet data session; and
sending the inspection parameters to another network node.
2. The method of claim 1 , further comprising registering, with a service discovery function, a capability to provide the one or more inspection parameters as an offered service.
3. The method of claim 1 , wherein sending the inspection parameters to another network node comprises:
receiving a request for the inspection parameters from a consumer node; and
sending, responsive to the request, the inspection parameters to the consumer node.
4. The method of claim 3 , wherein the consumer node comprises a data analytics function.
5. The method of claim 1 , wherein the inspection parameters include one or more of:
an average number of packets inspected to classify a traffic flow;
a number of packet detection rules (PDR) to apply to the traffic flow;
an average number of PDRs applied to classify a traffic flow;
a number of service functions to be traversed according to a policy;
an average time needed to classify a traffic flow; or
a type of analysis used for classifying the traffic flow.
6. The method of claim 1 , wherein the inspection parameters comprise per subscriber inspection parameters for one or more subscribers.
7. The method of claim 1 , wherein the inspection parameters comprise per service inspection parameters for one or more services.
8. The method of claim 1 , wherein the inspection parameters comprise per service inspection parameters for each service for each of one or more subscribers.
9. The method of claim 1 , wherein the inspection parameters comprise per subscriber type inspection parameters for each of one or more subscriber types.
10. A method implemented by a network node in a wireless communication network to support selection of a user plane function (UPF) by a session management function (SMF) for a packet data session, the method comprising:
receiving one or more inspection parameters associated with packet inspection for a subscriber, a service, or both;
computing a per subscriber and/or or per service computational footprint associated with packet inspection for the subscriber, the service, or both, based on the received packet inspection parameters, wherein the computational footprint is indicative of the computational resources needed to perform packet inspection for one or more traffic flows included in the packet data session; and
making the computational footprint for the subscriber, the service, or both, available to a session management function for use in UPF selection.
11. The method of claim 10 , wherein receiving one or more inspection parameters associated with packet inspection for a subscriber, a service, or both, comprises:
sending a service request for the inspection parameters to a producer node; and
receiving the inspection parameters from the producer node responsive to the request.
12. The method of claim 11 wherein the producer node comprises a UPF.
13. The method of claim 10 , wherein the inspection parameters include one or more of:
an average number of packets inspected to classify a traffic flow;
a number of packet detection rules (PDR) to apply to the traffic flow;
an average number of PDRs applied to classify a traffic flow;
a number of service functions to be traversed according to a policy;
an average time needed to classify a traffic flow; or
a type of analysis used for classifying the traffic flow.
14. The method of claim 10 , wherein the inspection parameters comprise per subscriber inspection parameters for one or more subscribers.
15. The method of claim 10 , wherein the inspection parameters comprise per service inspection parameters for one or more services.
16. The method of claim 10 , wherein the inspection parameters comprise per service inspection parameters for each service for each of one or more subscribers.
17. The method of claim 10 , wherein the inspection parameters comprise per subscriber type inspection parameters for each of one or more subscriber types.
18. The method of claim 10 , wherein making the computational footprint for the subscriber, the service, or both, available to a session management function for use in UPF selection comprises storing the computational footprint in a data repository.
19. A method implemented by a session management node in a wireless communication network of selecting a user plane function (UPF) for a packet data session, the method comprising:
receiving a request to create a packet data session for a user equipment (UE);
responsive to the request to create a packet data session, obtaining session management subscription data including a per subscriber and/or or per service computational footprint indicative of the computational resources needed to perform packet inspection for one or more traffic flows included in the packet data session; and
selecting a user plane function for the packet data session based at least in part on the computational footprint for packet inspection.
20. The method of claim 19 , wherein obtaining the session management data comprises retrieving the session management data from a subscriber database.
21. The method of claim 19 , wherein the computational footprint comprises a CPU demand, a memory demand, or both, associated with the user equipment.
22. The method of claim 19 , wherein the computational footprint comprises a CPU demand, a memory demand, or both, for each of one or more services associated with the packet data session.
23. The method of claim 19 , wherein the computational footprint comprises a CPU demand, a memory demand, or both, for each of one or more services associated with the packet data session for the UE.
24. The method of claim 1 , further comprising:
receiving, from a user equipment (UE), a request to establish the packet data session; and
selecting, responsive to the request, a UPF for the packet data session based on the computational footprint.