Network load balancing using multicast or packet replication

• US 9,762,507 B1
• Filed: 12/12/2013
• Issued: 09/12/2017
• Est. Priority Date: 03/14/2013
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Gandham, Shashidhar Rao, Contavalli, Carlo
• Primary Examiner(s): Rahman, Mahfuzur
• Assistant Examiner(s): Victoria, Narciso
• Application Number:
  US14/104,413
• Time in Patent Office: 1,370 Days
• Field of Search: 726 22
• CPC Class Codes: H04L 47/806 Broadcast or multicast traffic
• CPC Class Codes: H04L 49/15 Interconnection of switching modules
• CPC Class Codes: H04L 63/101 Access control lists [ACL]
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers

Abstract

Network load balancing may be performed in a cluster by deploying multicasting or replication. In particular, one or more multicast trees may be deployed within the cluster. Packets may be multicast towards the hosts over the multicast tree(s) to reach the hosts in a pool of hosts associated with a given Virtual Internet Protocol (“VIP”) address. Packets may be filtered before the packets reach the hosts to ensure that each host only receives packets for the session the host is handling. The filtering may be performed at various levels of the multicast tree. Replication may be deployed instead of multicasting to realize the same network load balancing.

Claims

1. A method of performing network load balancing in a cluster having switching nodes and hosts, comprising:

defining a first multicast tree in the cluster for a service, wherein the first multicast tree extends from an initial one of the switching nodes to a first pool of the hosts;

defining a second multicast tree for the service in the cluster, wherein the second multicast tree extends from a selected one of the switching nodes to a second pool of the hosts;

multicasting over the first multicast tree so that packets reach hosts in the first pool of hosts hosting sessions associated with the respective packets, wherein the packets are incoming packets for the service to the initial one of the switching nodes;

multicasting the incoming packets for the service over the second multicast tree so that packets reach hosts in the second pool of hosts hosting sessions associated with the respective incoming packets;

filtering out at least one of the packets at one of the switching nodes such that each host in the first and second pools of the hosts only receives packets associated with sessions the respective hosts are hosting, wherein the switching nodes are arranged in multiple layers and the filtering occurs at an aggregation switching node within the multiple layers of switching nodes; and

processing the packets with the first pool of hosts and the second pool of hosts.

2. The method of claim 1, further comprising:

splitting packet traffic for the service among the first multicast tree and the second multicast tree.

3. The method of claim 1, wherein the selected one of the switching nodes is the initial one of the switching nodes.

4. The method of claim 1, wherein the second pool of hosts is the first pool of hosts.

5. The method of claim 4, further comprising:

defining a third multicast tree in the cluster for a service, wherein the multicast tree extends from an initial one of the switching nodes to the first pool of the hosts;

splitting packet traffic among the first multicast tree, the second multicast tree and the third multicast tree.

6. The method of claim 1, further comprising:

implementing one or more access control lists (ACLs) at one or more of the switching nodes; and

filtering out attack traffic using the ACLs.

7. The method of claim 1, wherein the first multicast tree and the second multicast tree are disjoint.

8. A method of performing network load balancing in a cluster having switching nodes and hosts, comprising:

defining a first multicast tree in the cluster for a service, wherein the first multicast tree extends from an initial one of the switching nodes to a pool of the hosts;

receiving an incoming packet at the initial one of the switching nodes;

replicating the incoming packet at the initial one of the switching nodes;

forwarding the incoming packet from the initial one of the switching nodes to a first set of the switching nodes in the first multicast tree;

forwarding the replicated packet from the initial one of the switching nodes to a second set of the switching nodes in the first multicast tree, the second set of the switching nodes disjoint from the first set of the switching nodes;

filtering out at least one of the incoming packet and the replicated packet at one of the switching nodes such that each host in the pool of the hosts only receives packets associated with sessions the respective hosts are hosting, wherein the switching nodes are arranged in multiple layers and the filtering occurs at an aggregation switching node within the multiple layers of switching nodes; and

processing, at hosts within the pool of hosts, at least one of the incoming packet and the replicated packet that arrives at the hosts.

9. The method of claim 8, further comprising:

replicating at least a portion of the incomings packets at one or more of the first set of the switching nodes;

or

replicating at least a portion of the replicated incomings packets at one or more of the second set of the switching nodes.

10. The method of claim 9, further comprising:

implementing one or more access control lists (ACLs) at one or more of the switching nodes; and

filtering out attack traffic using the ACLs.

11. The method of claim 9, further comprising:

defining a second multicast tree in the cluster for the service, wherein;

the second multicast tree extends from the initial one of the switching nodes to the pool of hosts, and

the first multicast tree and the second multicast tree are disjoint;

replicating incomings packets received at the initial switching node;

forwarding the incoming packets to a first set of switching nodes in the second multicast tree; and

forwarding the replicated packets to a second set of switching nodes in the second multicast tree so that packets reach the pool of hosts.

12. A system comprising:

a network controller comprising at least one computer processor configured to execute computer executable instructions, the computer executable instructions configured such that when executed they carry out a method comprising;

defining a first multicast tree in a cluster for a service, wherein the first multicast tree extends from an initial switching node in the cluster to a pool of hosts;

receiving an incoming packet at the initial one of the switching nodes;

replicating the incoming packet at the initial one of the switching nodes;

forwarding the incoming packet from the initial one of the switching nodes to a first set of switching nodes in the first multicast tree;

forwarding the replicated packet from the initial one of the switching nodes to a second set of the switching nodes in the first multicast tree, the second set of the switching nodes disjoint from the first set of the switching nodes;

filtering out at least one of the incoming packet and the replicated packet at one of the switching nodes such that each host in the pool of the hosts only receives packets associated with sessions the respective hosts are hosting, wherein the switching nodes are arranged in multiple layers and the filtering occurs at an aggregation switching node within the multiple layers of switching nodes; and

processing, at hosts within the pool of hosts, at least one of the incoming packet and the replicated packet that arrives at the hosts.

13. The system of claim 12, wherein the network controller further executes instructions for:

replicating at least a portion of the incomings packets at one or more of the first set of switching nodes;

or

replicating at least a portion of the replicated incomings packets at one or more of the second set of switching nodes.

14. The system of claim 12, wherein the network controller further executes instructions for:

implementing one or more access control lists (ACLs) at one or more of the first set of switching nodes or the second set of switching nodes; and

filtering out attack traffic using the ACLs.

15. The system of claim 12, wherein the network controller further executes instructions for:

defining a second multicast tree in the cluster for the service, wherein;

the second multicast tree extends from the initial switching node in the cluster to the pool of hosts, and

the first multicast tree and the second multicast tree are disjoint;

replicating incomings packets received at the initial switching node; and

forwarding the incoming packets to a first set of switching nodes in the second multicast tree and the replicated packets to a second set of switching nodes in the second multicast tree so that packets reach the pool of hosts.

16. The system of claim 15, wherein the network controller further executes instructions for:

splitting network traffic for the service among the first multicast tree and the second multicast tree.