Granted Patent B2

US 11,671,371 · App. 16/941,877 · Granted Jun 6, 2023

Inventors: Anil Abraham (Karnataka, IN); Manikandan Kenyan (Saratoga, CA); Ashok Babu Thangamani (Bangalore, IN)

Assignee: Musarubra US LLC

H04L47/125H04L45/02H04L45/021H04L47/215

Abstract

Claims (37)

1. A method of a work node synchronously load balancing to a multi-node service having an expected maximum of n work nodes, comprising:

provisioning a flow table having m bucket groups, m≥1, the bucket groups comprising n slots each;

enumerating a static integer self-identification id 0 ;

initializing the flow table with id 0 in each slot;

performing a discovery iteration, comprising:

discovering a peer device;

enumerating a static integer identification id x for the peer device;

assigning id x to each slot corresponding to a home position for the peer device; and

load balancing slots not assigned to a home position according to a deterministic algorithm; and

discovering additional nodes and performing discovery iterations for the additional nodes.

2. The method of claim 1 , wherein the home position for a device is at a numerical offset within the bucket group that corresponds to id x .

3. The method of claim 1 , wherein the deterministic algorithm comprises cycling through static integer identifications of discovered devices in ascending numerical order.

4. One or more tangible, non-transitory computer-readable storage media comprising instructions to:

join a multi-node service cluster of up to n nodes;

receive a static numeric identifier;

fully populate a flow table with the static numeric identifier, wherein the flow table has n slots;

discover a new node in the cluster;

receive a second static numeric identifier for the new node;

assign the second static numeric identifier to one or more home positions for the second numeric identifier within the flow table, wherein the one or more home positions are reserved for the second static numeric identifier;

load balance remaining positions within the flow table according to a deterministic method; and

upon discovering another new node in the cluster, repeat the receive-assign-load balance sequence for the other new node.

5. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein load balancing the remaining positions is according to a deterministic algorithm.

6. The one or more tangible, non-transitory computer-readable storage media of claim 5 , wherein the deterministic algorithm comprises sequentially filling non-home positions with sequentially numeric identifiers of discovered nodes in ascending numerical order.

7. The one or more tangible, non-transitory computer-readable storage media of claim 6 , wherein the deterministic algorithm further comprises cycling back to a smallest numeric identifier.

8. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the home position for a device is at a numerical offset within a bucket group that corresponds to a numeric identifier for that node.

9. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein load balancing remaining positions within the flow table comprises applying a deterministic algorithm, wherein the deterministic algorithm comprises cycling through static integer identifications of discovered devices.

10. The one or more tangible, non-transitory computer-readable storage media of claim 9 , wherein the deterministic algorithm comprises cycling through static integer identifications of discovered devices in ascending numerical order.

11. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the instructions are further to provide a network security service.

12. The one or more tangible, non-transitory computer-readable storage media of claim 11 , wherein the network security service is a stateful flow-based service.

13. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the new nodes are substantially identical.

14. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the new nodes share a common trunk.

15. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the new nodes are arranged in a dual stack, comprising a primary stack and a failover stack.

16. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein n=4.

17. The one or more tangible, non-transitory computer-readable storage media of claim 4 , wherein the nodes provide deep packet inspection.

18. A network service cluster comprising at least one hardware platform comprising a processor and a memory, and sharing a common trunk comprising a plurality of service nodes, wherein the service nodes are programmed to maintain independent flow tables for load balancing, wherein the independent flow tables have a number of slots corresponding to a number of nodes in the network service cluster and are to converge over time according to a deterministic algorithm wherein each node initially fully populates its flow table with its own numeric identifier, discovers other nodes, and for each discovered node assigns a numeric identifier for the discovered node to a home position for that numeric identifier, and then load balances remaining positions in the flow table between discovered nodes.

19. The network service cluster of claim 18 , wherein the home position for a device is at a numerical offset within a bucket group that corresponds to the device's identification.

20. The network service cluster of claim 18 , wherein the deterministic algorithm comprises cycling through static integer identifications of discovered devices.

Assignments (14)

Continuity (1)