ENCODING A PAYLOAD HASH IN THE DA-MAC TO FACILITATE ELASTIC CHAINING OF PACKET PROCESSING ELEMENTS

• US 20140341029A1
• Filed: 05/20/2013
• Published: 11/20/2014
• Est. Priority Date: 05/20/2013
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Allan, David Ian, Gray, Eric Ward, Halpern, Joel, Mansfield, Scott Andrew
• Granted Patent: US 9,160,666 B2
• Time in Patent Office: Days
• US Class Current: 370/235
• CPC Class Codes: H04L 41/12 Discovery or management of network topologies
• CPC Class Codes: H04L 45/02 Topology update or discovery
• CPC Class Codes: H04L 45/0377 for service chaining
• CPC Class Codes: H04L 45/38 Flow based routing
• CPC Class Codes: H04L 45/7453 using hashing
• CPC Class Codes: H04L 47/125 by balancing the load, e.g. traffic engineering
• CPC Class Codes: H04L 49/15 Interconnection of switching modules

Abstract

A method is implemented in a network element of a service network. The network element executes a packet processing element (PPE) of a plurality of PPEs, where each PPE in the plurality of PPEs executes a stage of packet processing for the service network and where the plurality of PPEs are connected to one another by a plurality of switch fabrics. The PPEs self-select a subset of a set of equivalent service chains to service with each service chain defining a subset and sequence of the plurality of PPEs. Each PPE self-selects the subset of equivalent service chains to process based upon knowledge of the plurality of PPEs servicing that stage of the full set of service chains such that there is spreading of load across all available PPEs in that stage. There is no single point of failure and minimal reassignment of PPEs for a set of equivalent data flows traversing the plurality of PPEs of the service network for changes in topography of the plurality of PPEs in the service network.

Claims

1. A method implemented in a network element in a service network, the network element to execute a packet processing element (PPE) of a plurality of PPEs, where each PPE in the plurality of PPEs executes a stage of packet processing for the service network, where the plurality of PPEs are connected to one another by a plurality of switch fabrics, the PPE to self-select a subset of a set of equivalent service chains to service, each service chain defining a subset and sequence of the plurality of PPEs, the PPE to self-select the subset of equivalent service chains to process based upon knowledge of the plurality of PPEs servicing that stage of the set of service chains such that there is spreading of load across all available PPEs in that stage, there is no single point of failure and minimal reassignment of PPEs for a set of equivalent data flows traversing the plurality of PPEs of the service network from changes in topography of the plurality of PPEs in the service network, the method comprising the steps of:

detecting a change in the topology of the plurality of PPEs;

generating a next load balancing (LB) table to adjust service chain routes for the changed topology, the generating using a shared transform algorithm to replace a current LB table, the next LB table and the current LB table identifying next hops for all service chains in the service network;

sharing the current LB table and the next LB table with the plurality of PPEs in the service network;

transitioning the PPE to the next LB table by replacing the current LB table; and

programming an upstream switch fabric to forward to itself a specific subset of frames selected by the PPE for processing according to the next LB table.

2. The method of claim 1, wherein the sharing of the current LB table and the next LB table comprises the steps of:

generating a digest of the current LB table and the next LB table where the next LB table has been adjusted in accordance to a detected change, and

broadcasting the digest to the plurality of PPEs.

3. The method of claim 1, further comprising the steps of:

confirming quiescence of current flow hash selections of the PPE before the transitioning to the next LB table.

4. The method of claim 1, further comprising the steps of:

confirming synchronization of the current LB table and the next LB table amongst the plurality of PPEs before transitioning to the next LB table.

5. The method of claim 1, further comprises the steps of:

receiving a frame from the upstream switching fabric;

checking whether a data flow identifier of the frame is selected by the PPE; and

discarding the frame in response to a failure to match the flow hash with the PPE.

6. The method of claim 5, further comprises the steps of:

forwarding the frame to a downstream switching fabric after processing by the PPE, preserving the data flow identifier in the frame.

7. The method of claim 6, wherein the data flow identifier is encoded in a destination media access control address (DA MAC) of the frame at ingress to the service network.

8. The method of claim 6, wherein a service chain identifier is encoded in a virtual local area network (VLAN) such that multiple service chains can self-identifier within a single matching fabric.

9. The method of claim 1, further comprising the steps of:

receiving a data packet;

determining a VLAN identifier for the service chain for the data packet;

applying a shared hashing function to header information of data packet to create the flow hash;

encoding the flow hash as destination media access control address (DA MAC) of frame corresponding to received data packet to be forwarded the service chain; and

forwarding the frame to the switching fabric configured by the PPE.

10. The method of claim 1, further comprising the step of:

detecting an operations administration and management (OAM) frame having the data flow identifier and forwarding the OAM frame as a standard frame with the data flow identifier.

11. A network element in a service network, the network element to execute a packet processing element (PPE) of a plurality of PPEs, where each PPE in the plurality of PPEs executes a stage of packet processing for the service network, where the plurality of PPEs are connected to one another by a plurality of switch fabrics, the PPE to self-select a subset of a set of equivalent service chains to service, each service chain defining a subset and sequence of the plurality of PPEs, the PPE to self-select the subset of equivalent service chains to process based upon proper knowledge of the plurality of PPEs servicing that stage of the set of service chains such that there is spreading of load across all available PPEs in that stage, there is no single point of failure, and minimal reassignment of PPEs for a set of equivalent data flows traversing the plurality of PPEs of the service network from changes in topography of the plurality of PPEs in the service network, the network element comprising:

a memory to provide data storage for a network processor;

a network processor coupled to the memory, the network processor configured to execute the PPE, the PPE including a load balancing (LB) synchronization module, an LB table management module, an LB generation module and a switch configuration module,

the LB synchronization module configured to detect a change in the topology of the plurality of PPEs and share a current LB table and next LB table with the plurality of PPEs in the service network,

the LB table management module configured to transition the PPE to the next LB table by replacing the current LB table,

the LB generation module configured to generate the next load balancing (LB) table to adjust service chain routes for the changed topology, the generation using a shared transform algorithm to replace the current LB table, the next LB table and the current LB table identifying next hops for all service chains in the service network, and

the switch configuration module configured to program an upstream switch fabric to forward to itself a specific subset of frames selected by the PPE for processing according to the next LB table.

12. The network element of claim 11, wherein the LB synchronization module is further configured to generate a digest of the current LB table and the next LB table, where the next LB table has been adjusted in accordance to a detected change, and broadcast the digest to the plurality of PPEs.

13. The network element of claim 11, wherein the LB table management module is further configured to confirm quiescence of current flow hash selections of the PPE before the transitioning to the next LB table.

14. The network element of claim 11, wherein the LB synchronization module is further configured to confirm synchronization of the current LB table and the next LB table amongst the plurality of PPEs before transitioning to the next LB table.

15. The network element of claim 11, wherein the PPE further comprises:

a frame checker configured to receive a frame from the upstream switching fabric and check whether a data flow identifier of the frame is selected by the PPE, and discard the frame in response to a failure to match the flow hash with the PPE.

16. The network element of claim 15, further comprising:

a forwarding module configured to forward the frame to a downstream switching fabric after processing by the PPE, preserving the data flow identifier in the frame.

17. The network element of claim 16, wherein the data flow identifier is encoded in a destination media access control address (DA MAC) of the frame at ingress to the service network.

18. The network element of claim 17, wherein a service chain identifier is encoded in a virtual local area network (VLAN) such that multiple service chains can self-identify within a single switching fabric.

19. The network element of claim 11, further comprising:

a service chain identification module configured to receive a data packet and determine a VLAN identifier (VID) for the service chain for the data packet;

a data flow identifier generation module configured to apply a shared hashing function to header information of data packet to create the flow hash and encode the flow hash as destination media access control address (DA MAC) of frame corresponding to received data packet to be forwarded the service chain; and

a service chain forwarding module configured to forward the frame to the switching fabric configured by the PPE.