Enhanced network system through the combination of network objects

• US 20060209718A1
• Filed: 03/17/2005
• Published: 09/21/2006
• Est. Priority Date: 03/17/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Garg, Pankaj, Kinsey, Jeffrey Brian, Jain, Neel K.
• Granted Patent: US 7,797,460 B2
• Time in Patent Office: Days
• US Class Current: 370/254
• CPC Class Codes: G06F 2009/45595 Network integration; Enabling network access in virtual machine instances
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects
• CPC Class Codes: H04L 2212/00 Encapsulation of packets
• CPC Class Codes: Y10S 370/903 Osi compliant network

Abstract

A system and method for reducing the cost of framing network packets in a virtual machine environment combines certain network objects to eliminate the cost of fully framing packets between the combined objects. In a virtual environment, for example, this allows a virtual NIC in one partition to send only data to a network provider partition and to rely on the latter to frame and route the data. A source network object, destination network object, or an intermediate network object may enable a separate intermediate network object to frame its data so that the network system may take advantage of offload framing when one or more of the intermediate network object'"'"'s ports has the capacity to offload framing. Alternatively, the intermediate network object may perform the offload itself. Also, the source or destination network object may query the intermediate network object for properties (such as the MTU of the network), or the intermediate network object may push its properties to the source and/or destination network objects through a back control data communications channel. The intermediate network object may further reconfigure the VM network for the purpose of efficiency by reconfiguring the network tree into a graph for increased performance for traffic moving between the nodes.

Claims

1. A virtual machine system comprising at least a source network object, an intermediate network object, and a destination network object, comprising:

a network channel that connects said source network object, said intermediate network object, and said destination network object, said network channel being configured to send framed data packets among said source network object, said intermediate network object, and said destination network object; and

a control channel this is separate from said network channel and connects at least two of said source network object, said intermediate network object, and said destination network object, said control channel being configured to send control data messages between said at least two network objects.

2. A virtual machine system as in claim 1, wherein said control channel comprises a software connection between respective network objects.

3. A virtual machine system as in claim 1, wherein said control channel comprises a packet bus connection between respective network objects.

4. A virtual machine system as in claim 1, wherein said control data messages comprise a sub-protocol of a standard protocol stack for the framed data packets.

5. A virtual machine system as in claim 1, wherein said control data messages include routing/topology data for reconfiguring the respective network objects to change a routing of said framed data packets between at least said intermediate network object and said destination network object.

6. A virtual machine system as in claim 1, wherein said control data messages enable respective network objects to setup a path for the sending of not fully framed data packets, or data packets that are inappropriate for transmission, therebetween.

7. A virtual machine system as in claim 6, wherein said control data messages include TCP/IP checksums.

8. A virtual machine system as in claim 6, wherein said control data message includes a request that a TCP/IP offload engine (TOE) connection be established.

9. A virtual machine system as in claim 8, wherein a network object receiving said TOE connection request looks up a port corresponding to the network address of the TOE connection request in a routing table to determine that the particular network address is available via an adjacent network object and establishes a communication link between the network object that sent said TOE connection request and the determined adjacent network object by forwarding said received TOE connection request to the adjacent network object.

10. A virtual machine system as in claim 8, wherein if a network object receiving said TOE connection request determines that the TOE connection request should not be forwarded, said network object handles the TOE connection request locally.

11. A virtual machine system as in claim 10, wherein said network object receiving said TOE connection request determines that said TOE connection request should not be forwarded when the network object receiving the TOE connection request controls a physical NIC which can perform the TOE connection request in hardware or the network object receiving the TOE connection request is unable to find an adjacent network object which could handle the TOE connection request.

12. A virtual machine system as in claim 6, wherein said control data messages enable a network object besides the source network object to break apart a partially or fully framed data packet using a large send offload.

13. A virtual machine system as in claim 6, wherein said control data messages enable a network object besides the source network object to frame said data packets with meta data including routing information for each data packet and information on what each data packet contains for transmission to another network object via said network channel.

14. A virtual machine system as in claim 13, wherein said network object besides the source network object offloads a received message to framing hardware that frames said message into TCP/IP packets.

15. A virtual machine system as in claim 1, wherein said control data messages include at least one network property of a network object.

16. A virtual machine system as in claim 15, wherein said network property comprises a maximum transmission unit (MTU) size for a communication path between respective network objects.

17. A virtual machine system comprising at least a source network object, a destination network object, and an intermediate network object between said source network object and said destination network object, each of said network objects having network connections, wherein the intermediate network object provides routing data to the source network object, said routing data identifying at least one of a next intermediate network object and the destination network object in a network topology of said virtual machine system, said source network object using said routing data to create a communications path directly to said destination network object or said next intermediate network object so as to exclude said intermediate network object from said communications path.

18. The virtual machine system of claim 17, wherein said next intermediate network object provides further routing data to the source network object, said further routing data identifying at least one of an additional intermediate network object and the destination network object in said network topology of said virtual machine system, said source network object using said further routing data to create a communications path directly to said destination network object or said additional intermediate network object so as to exclude said intermediate network object and said next intermediate network object from said communications path.

19. The virtual machine system of claim 17, further comprising an additional intermediate network object between said source network object and said destination network object, wherein when the source network object determines that said intermediate network object and said additional network object should be combined said source network object sends a message to both said intermediate network object and said additional intermediate network object instructing said intermediate network object and said additional intermediate network object to create a connection to a particular network address, whereby subsequent messages between said intermediate network object and said additional intermediate network object may pass through each said connection.

20. A method of communicating data between respective network objects in a virtualized computer system, comprising:

determining a topography of a communication path between a source network object and a destination network object in said virtualized computer system, said communication path including at least one intermediate network object;

said at least one intermediate network object providing routing data to said source network object, said routing data identifying at least one of (1) an address of one or more subsequent intermediate network objects in said communication path and (2) a topography of a communication path from said at least one intermediate network object to said destination network object; and

establishing a communication link between said source network object and either said one or more subsequent intermediate network objects or said destination network object using said routing data so as to eliminate said at least one intermediate network object from a routing of data between said source network object and said destination network object.

21. A method as in claim 20, wherein said routing data providing step comprises reading routing tables of said at least one intermediate network object and providing routing data from said routing tables to said source network object.

22. A method as in claim 20, wherein said communication link establishing step comprises creating a communication link to either said one or more subsequent intermediate network objects or said destination network object using routing data stored in said source network object wherein said routing data excludes at least said at least one intermediate network object from the communication link.

23. A method as in claim 20, wherein said communication link is separate from said communication path and is configured to route control data including said routing data to said source network object.