Load balancing in a logical pipeline

• US 9,059,999 B2
• Filed: 02/01/2013
• Issued: 06/16/2015
• Est. Priority Date: 08/17/2011
• Status: Active Grant

• Current Assignee: Nicira Incorporated (Broadcom, Inc.)
• Original Assignee: Nicira Incorporated (Broadcom, Inc.)
• Inventors: Koponen, Teemu, Zhang, Ronghua
• Primary Examiner(s): Chu, Wutchung
• Application Number:
  US13/757,681
• Publication Number:
  US 20130148505A1
• Time in Patent Office: 865 Days
• Field of Search: 370/352, 370386-392, 370/235
• US Class Current: 1/1
• CPC Class Codes: H04L 41/0803 Configuration setting
• CPC Class Codes: H04L 45/04 Interdomain routing, e.g. hierarchical routing
• CPC Class Codes: H04L 45/54 Organization of routing tables
• CPC Class Codes: H04L 45/74 Address processing for routing
• CPC Class Codes: H04L 47/12 Avoiding congestion; Recovering from congestion
• CPC Class Codes: H04L 47/125 by balancing the load, e.g. traffic engineering
• CPC Class Codes: H04L 61/103 across network layers, e.g. resolution of network layer into physical layer addresses or address resolution protocol [ARP]
• CPC Class Codes: H04L 61/256 NAT traversal
• CPC Class Codes: H04L 61/2592 using tunnelling or encapsulation

Abstract

A non-transitory machine readable medium storing a program that configures a managed forwarding element to perform logical L2 switching and L3 routing is described. The program generates a first set of flow entries for configuring the first managed forwarding element to perform (1) a first logical L2 processing for a first logical L2 domain, (2) a logical L3 processing, (3) a load balancing processing to select a second managed forwarding element from a plurality of managed forwarding elements to which to forward packets and (4) a logical ingress L2 processing for a second logical L2 domain on the packets. The program generates a second set of flow entries for configuring the second managed forwarding element to perform a second logical L2 processing for a second logical L2 domain on the packets.

Claims

1. A non-transitory machine readable medium storing a program which when executed by at least one processing unit configures a plurality of managed forwarding elements (MFEs) to implement a logical L3 router and a plurality of logical L2 switches that each logically couple to a plurality of machines, the program comprising sets of instructions for:

for each MFE of a set of MFEs that implement a particular logical L2 switch, generating a first set of data records for configuring the MFE to install a first set of flow entries that implement the particular logical L2 switch and the logical L3 router for processing packets sent to a network address that logically couples to the particular logical L2 switch; and

for each MFE of the set of MFEs, generating a second set of data records for configuring the MFE to install a second set of flow entries that implement load balancing processing on a subset of packets sent to the network address that logically couples to the particular logical L2 switch, the second set of flow entries specifying to balance the subset of packets across a plurality of machines that logically couple to the particular logical L2switch and are physically coupled to a plurality of different MFEs.

2. The machine readable medium of claim 1, wherein the program further comprises a set of instructions for generating, for each MFE of the set of MFEs, a third set of data records for configuring the MFE to install a third set of flow entries that implement source network address translation (NAT) processing on packets sent from a second particular logical L2 switch to the particular logical L2 switch before performing the load balancing processing on said packets.

3. The machine readable medium of claim 1, wherein the plurality of machines that logically couple to the particular logical L2 switch comprise a plurality of virtual machines that provide a same service.

4. The machine readable medium of claim 1, wherein each of the MFEs in the set of MFEs is configured in a different host, wherein the program further comprises a set of instructions for configuring, in each of the hosts, a load balancing daemon to select a machine from the plurality of machines that logically couple to the particular logical L2switch based on current workload on each of the plurality of machines that logically couple to the particular logical L2 switch.

5. The machine readable medium of claim 4, wherein the set of instructions for configuring a particular load balancing daemon operating in a particular host with a particular MFE comprises a set of instructions for configuring the particular load balancing daemon to send an address of a selected machine that logically couples to the particular logical L2switch to the particular MFE.

6. The machine readable medium of claim 4, wherein the set of flow entries installed on the particular MFE for performing the load balancing processing specifies for the sarticular MFE to request from the load balancing daemon a selection of one of the machines that logically couples to the particular logical L2 switch.

7. The machine readable medium of claim 1, wherein the program further comprises a set of instructions for configuring each MFE of the set of MFEs to establish a tunnel with each of the other MFEs of the set of MFEs.

8. For a network controller that manages a network, a method for configuring a plurality of network elements in the network to implement a logical L2 domain, to which a plurality of machines couple, and a logical router, the method comprising:

configuring a plurality of edge managed forwarding elements (MFEs) to perform logical switching for the logical L2 domain, the plurality of edge MFEs operating in host machines on which machines that logically couple to the logical L2 domain reside;

configuring a separate managed forwarding element that operates in a host machine as a second-level managed forwarding element to perform logical switchin for the logical L2domain; and

configuring a container in the host machine to operate as a router for routing packets sent to and from the logical L2 domain and for performing load balancing on packets sent to the logical L2 domain by configuring the container to select a destination machine from the plurality of machines that logically couple to the logical L2domain, wherein the plurality of machines provide identical services, wherein the second-level MFE is configured to send to the container for routing packets addressed to a particular destination address that logically couples to the logical L2 domain.

9. The method of claim 8, wherein the second-level managed forwarding element facilitates communications between the edge MFEs, wherein the edge MFEs interface directly with, the plurality of machines that logically couple to the logical L2domain, wherein the second-level MFE does not directly interface with the plurality of machines.

10. The method of claim 8, wherein the container in the host machine is a namespace.

11. The method of claim 8 further comprising establishing tunnels between the second-level MFE and each of the edge MFEs.

12. The method of claim 8, wherein the edge MFEs are software switches and the machines are virtual machines.

13. For a network controller, a method for configuring a plurality of managed forwarding elements to implement a logical L3 router and a plurality of logical L2 switches, the method comprising:

for each MFE of a set of MFEs that implement a particular logical L2switch, generating a first set of data records for configuring the MFE to install a first set of flow entries that implement the particular logical L2 switch and the logical L3 router for processing packets sent to a network address that logically couples to the particular logical L2 switch; and

for each MFE of the set of MFEs, generating a second set of data records for configuring the MFE to install a second set of flow entries that implement load balancing processing on a subset of packets sent to the network address that logically couples to the particular logical L2 switch, the second set of flow entries specifying to balance the subset of packets across a plurality of machines that logically couple to the particular logical L2 switch and are physically coupled to a plurality of different MFEs.

14. The method of claim 13 further comprising generating, for each MFE of the set of MFEs, a third set of data records for configuring the MFE to install a third set of flow entries that implement source network address translation (NAT) processing on packets sent from a second particular logical L2 switch to the particular logical L2 switch before performing the load balancing processing on said packets.

15. The method of claim 13, wherein the plurality of machines that logically couple to the particular logical L2 switch comprise a plurality of virtual machines that provide a same service.

16. The method of claim 13, wherein each of the MFEs in the set of MFEs is configured in a different host, the method further comprising configuring, in each of the hosts, a load balancing daemon to select a machine from the plurality of machines that logically couple to the particular logical L2 switch based on current workload on each of the plurality of machines that logically couple to the particular logical L2switch.

17. The method of claim 16, wherein configuring a particular load balancing daemon operating in a particular host with a particular MFE comprises configuring the particular load balancing daemon to send an address of a selected machine to the particular MFE.

18. The method of claim 16, wherein the set of flow entries installed on the particular MFE for performing the load balancing processing specifies for the particular MFE to request from the load balancing daemon a selection of one of the machines that logically couples to the particular logical L2 switch.

19. The method of claim 13 further comprising configuring each MFE of the set of MFEs to establish a tunnel with each of the other MFEs of the set of MFEs.