Agile data center network architecture

• US 9,497,039 B2
• Filed: 10/14/2009
• Issued: 11/15/2016
• Est. Priority Date: 05/28/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Greenberg, Albert, Lahiri, Parantap, Maltz, David A., Patel, Parveen K., Sengupta, Sudipta, Jain, Navendu, Kim, Changhoon
• Primary Examiner(s): Christensen, Scott B
• Application Number:
  US12/578,608
• Publication Number:
  US 20100306408A1
• Time in Patent Office: 2,589 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: H04L 12/4633 Interconnection of networks using encapsulation techniques, e.g. tunneling
• CPC Class Codes: H04L 45/16 Multipoint routing
• CPC Class Codes: H04L 49/356 for storage area networks
• 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 61/45 Network directories; Name-to-address mapping

Abstract

This patent application relates to an agile network architecture that can be employed in data centers, among others. One implementation provides a virtual layer-2 network connecting machines of a layer-3 infrastructure.

Claims

1. A method of providing a virtual layer-2 network over a layer-3 infrastructure connecting a plurality of machines including a first machine and a second machine by assigning application addresses to individual machines of the plurality of machines and location addresses to components of the layer-3 infrastructure, the method comprising:

intercepting, by an agile agent associated with the first machine, a virtual layer-2 packet with an assigned application address of a destination server associated with the second machine, wherein individual machines of the plurality of machines are associated with different agile agents;

determining, by the agile agent associated with the first machine, that the destination server is in a defined server group for a service, the defined server group being stored in an agile directory service;

based on the destination server being in the defined server group, utilizing, by the agile agent, the agile directory service to retrieve an individual location address associated with the application address;

selecting, by the agile agent, a switch of the layer-3 infrastructure through which to send the virtual layer-2 packet to the individual location address, the switch being selected from a plurality of switches of the layer-3 infrastructure to provide load balancing;

encapsulating, by the agile agent, the virtual layer-2 packet in a layer-3 packet, wherein the layer-3 packet is assigned the individual location address of the switch of the layer-3 infrastructure; and

transmitting the layer-3 packet to the switch, wherein the switch is configured to decapsulate the encapsulated virtual layer-2 packet and transmit the decapsulated virtual layer-2 packet to the second machine over a physical network connection,

wherein, where the destination server is not in the defined server group, the agile directory service refuses to provide the individual location address.

2. The method of claim 1, further comprising using early turnaround paths between the individual machines of the plurality of machines.

3. The method of claim 1, wherein the agile agent comprises a combination of hardware located within layer-2 infrastructure associated with the first machine and software that executes in the layer-2 infrastructure associated with the first machine.

4. The method of claim 1, wherein the providing the virtual layer-2 network comprises providing multiple virtual layer-2 networks.

5. The method of claim 1, wherein the agile agent is located on a source server associated with the first machine or on a Top of Rack (ToR) switch associated with the first machine, wherein the ToR switch is not the switch.

6. The method of claim 1, wherein the selecting the switch further comprises randomly selecting an individual path of the layer-3 infrastructure between the first and second machines.

7. The method of claim 6, further comprising utilizing valiant load balancing to select the individual path.

8. The method of claim 6, further comprising:

monitoring, by the agile agent, at least one network performance parameter; and

reselecting the individual path responsive to a based upon values of the at least one network performance parameter.

9. A system, comprising:

a plurality of computing devices communicatively coupled via a physical network connection including a plurality of switches, individual computing devices associated with an application address;

an agile agent associated with a source computing device of the plurality of computing devices, the agile agent configured to;

intercept a packet with an assigned application address of a destination computing device of the plurality of computing devices,

determine whether the destination computing device is in a defined communication group of the source computing device, the defined communication group being stored in an agile directory service,

where the destination computing device is in the defined communication group, retrieve a location address of an individual switch associated with the assigned application address using the agile directory service,

select a different individual switch of the plurality of switches through which to send the packet to the location address, the different individual switch being selected to provide load balancing,

encapsulate the packet and assign the encapsulated packet the location address of the individual switch, and

transmit the encapsulated packet to the individual switch via the different individual switch, wherein upon receipt of the encapsulated packet the individual switch is configured to decapsulate the packet and transmit the decapsulated packet to the application address of the destination computing device over a physical network connection,

wherein, where the destination computing device is not in the defined communication group, the agile directory service refuses to provide the location address.

10. The system of claim 9, wherein the individual switch associated with the assigned application address is a Top of Rack (ToR) switch.

11. The system of claim 9, wherein the agile agent comprises a plurality of agile agents and wherein sub-sets of the agile agents and the plurality of computing devices are organized into a server rack and further comprising a network cage configured to be communicatively coupled to multiple server racks.

12. The system of claim 9, wherein the plurality of switches comprise intermediate switches and aggregation switches that are arranged in a pluggable container with at least some of the plurality of computing devices.

13. The system of claim 9, wherein the plurality of switches comprises a layer-3 infrastructure and encapsulating the packet further comprises encapsulating a virtual layer-2 packet with a layer-3 packet for transmission over the layer-3 infrastructure.

14. The system of claim 9, further comprising the agile directory service, wherein the agile directory service is configured to assign sub-sets of the plurality of switches to individual groups of computing devices.

15. The system of claim 9, further comprising the agile directory service, wherein the agile directory service is configured to assign specific application addresses and specific location addresses to individual customers and to specify which application addresses are allowed to communicate with one another.

16. The system of claim 9, further comprising the agile directory service, wherein the agile directory service is configured to map paths between the source computing device and the destination computing device via individual switches using anycast addresses or valiant load balancing.

17. The system of claim 9, wherein the agile agent comprises hardware located on a forwarding path of the packet, the hardware configured to intercept the packet, the forwarding path being located between the source computing device and the destination computing device, and the agile agent further comprises software instructions that execute in the forwarding path to cause the interception of the packet.

18. The system of claim 17, wherein the plurality of switches are further along the forwarding path toward the destination computing device than the agile agent, and the plurality of switches are aware of the location address but are not aware of the application address.

19. A server, comprising:

at least one processor for executing computer readable instructions; and

,

an agile agent that is executable by the at least one processor and configured to;

receive a packet for delivery to another server with an application address,

access an agile directory service to determine whether a defined server group of the server includes the another server, the defined server group being stored in the agile directory service, and

in a case where the defined server group includes the another server, utilize the agile directory service to retrieve a location address of a switch associated with the another server, encapsulate the packet for the delivery over a physical network connection via an intermediate switch selected from a plurality of intermediate switches to provide load balancing, wherein the encapsulated packet is assigned the location address to the switch associated with the another server, and wherein upon receiving the encapsulated packet the switch associated with the another server is configured to decapsulate the encapsulated packet and transmit the decapsulated packet to the another server with the application address over a physical network connection,

wherein, in cases where the defined server group does not include the another server, the agile directory service denies the location address.

20. The server of claim 19, wherein the agile agent is further configured to provide path control associated with the load balancing.

21. The server of claim 19, wherein the defined server group is stored by the agile directory service.

22. The server of claim 19, wherein the agile directory service stores a mapping of the application address to the location address.