System and method for supporting VM migration between subnets in a high performance computing environment
System and method for supporting virtual machine migration in a high performance computing environment. In accordance with an embodiment, when a virtual machine migrates from a first subnet to a second subnet, a gateway port at the first subnet can, upon receiving packets addressed to the migrated virtual machine, consult a memory lookup table to discover the virtual machine's new addresses. The gateway port can update the packet accordingly. The memory lookup table can be addressed based on a GUID of the virtual machine.
1. A method for supporting virtual machine migration in a high performance computing environment, comprising:
providing, at one or more computers, including one or more microprocessors
a network fabric, the network fabric comprising
a first subnet, the first subnet comprising a first set of network switches, the first set of network switches comprising a gateway switch, the gateway switch having access to a memory;
a second subnet, the second subnet comprising a second set of network switches;
a virtual machine attached to the first subnet via a first virtual function, the virtual machine being associated with a first plurality of addresses, the first plurality of addresses comprising at least a global unique identifier (GUID), a global identifier (GID), and a destination local identifier (DLID); and
a global fabric manager, the global fabric manager having administrative control within the first subnet and second subnet;
detaching the virtual machine from the first virtual function;
migrating the virtual machine from the first subnet to the second subnet, wherein after migration the virtual machine is associated with a second plurality of addresses, the second plurality of addresses comprising the GUID, a second GID, and a second DLID, wherein the second GID is different from the GID, and wherein the second DLID is different from the DLID;
attaching the virtual machine to the second subnet via a second virtual function;
adding a temporary entry to the memory, the temporary entry comprising a set of addresses comprising the first plurality of addresses and the second plurality of addresses of virtual machine;
after migrating the virtual machine, receiving, at the gateway switch of the first subnet, a packet addressed to the virtual machine at the first plurality of addresses;
performing a lookup within the memory to find the second plurality of addresses of the virtual machine; and
updating a header of the received packet with the second plurality of addresses of the virtual machine.
2. The method of claim 1 , wherein performing the lookup comprises:
looking up the second plurality of addresses based upon the GUID of the migrated virtual machine.
3. The method of claim 2 , wherein the GUID of the virtual machine remains unchanged throughout the migration.
4. The method of claim 3 , wherein the memory comprises a ternary content-addressable memory (TCAM).
5. The method of claim 1 , further comprising:
removing the temporary entry from the memory after a period of time.
6. The method of claim 5 , wherein the period of time is determined based upon a length of time to update the network fabric with the second plurality of addresses.
7. A system for supporting virtual machine migration in a high performance computing environment, the system comprising:
a computer environment comprising:
a network fabric, the network fabric comprising
a first subnet, the first subnet comprising a first set of network switches, the first set of network switches comprising a gateway switch, the gateway switch having access to a memory;
a second subnet, the second subnet comprising a second set of network switches;
a virtual machine attached to the first subnet via a first virtual function, the virtual machine being associated with a first plurality of addresses, the first plurality of addresses comprising at least a global unique identifier (GUID), a global identifier (GID), and a destination local identifier (DLID); and
a global fabric manager, the global fabric manager having administrative control within the first subnet and second subnet;
wherein the virtual machine is configured to detach from the first virtual function;
wherein the virtual machine is migrated from the first subnet to the second subnet, wherein after migration the virtual machine is associated with a second plurality of addresses, the second plurality of addresses comprising the GUID, a second GID, and a second DLID, wherein the second GID is different from the GID, and wherein the second DLID is different from the DLID;
wherein the virtual machine is attached to the second subnet via a second virtual function;
wherein, upon the migration, a temporary entry to the memory is added, the temporary entry comprising a set of addresses comprising the first plurality of addresses and the second plurality of addresses of virtual machine;
wherein upon migrating the virtual machine, the gateway switch of the first subnet receives a packet addressed to the virtual machine at the first plurality of addresses;
wherein upon receiving the packet, the gateway switch performs a lookup within the memory to find the second plurality of addresses of the virtual machine; and
wherein a header of the received packet is updated with the second plurality of addresses of the virtual machine.
8. The system of claim 7 , wherein performing the lookup comprises:
looking up the second plurality of addresses based upon the GUID of the migrated virtual machine.
9. The system of claim 8 , wherein the GUID of the virtual machine remains unchanged throughout the migration.
10. The system of claim 9 , wherein the memory comprises a ternary content-addressable memory (TCAM).
11. The system of claim 7 , wherein the temporary entry to the memory is removed after a period of time.
12. The system of claim 11 , wherein the period of time is determined based upon a length of time to update the network fabric with the second plurality of addresses.
13. A non-transitory machine readable medium, including instructions stored thereon for supporting virtual machine migration in a high performance computing environment, which when read and executed by one or more computers caused the one or more computers to perform steps comprising:
providing, at one or more computers, including one or more microprocessors
a network fabric, the network fabric comprising
a first subnet, the first subnet comprising a first set of network switches, the first set of network switches comprising a gateway switch, the gateway switch having access to a memory;
a second subnet, the second subnet comprising a second set of network switches;
a virtual machine attached to the first subnet via a first virtual function, the virtual machine being associated with a first plurality of addresses, the first plurality of addresses comprising at least a global unique identifier (GUID), a global identifier (GID), and a destination local identifier (DLID); and
a global fabric manager, the global fabric manager having administrative control within the first subnet and second subnet;
detaching the virtual machine from the first virtual function;
migrating the virtual machine from the first subnet to the second subnet, wherein after migration the virtual machine is associated with a second plurality of addresses, the second plurality of addresses comprising the GUID, a second GID, and a second DLID, wherein the second GID is different from the GID, and wherein the second DLID is different from the DLID;
attaching the virtual machine to the second subnet via a second virtual function;
adding a temporary entry to the memory, the temporary entry comprising a set of addresses comprising the first plurality of addresses and the second plurality of addresses of virtual machine;
after migrating the virtual machine, receiving, at the gateway switch of the first subnet, a packet addressed to the virtual machine at the first plurality of addresses;
performing a lookup within the memory to find the second plurality of addresses of the virtual machine; and
updating a header of the received packet with the second plurality of addresses of the virtual machine.
14. The non-transitory machine readable medium of claim 13 , wherein performing the lookup comprises:
looking up the second plurality of addresses based upon the GUID of the migrated virtual machine.
15. The non-transitory machine readable medium of claim 14 , wherein the GUID of the virtual machine remains unchanged throughout the migration.
16. The non-transitory machine readable medium of claim 15 , wherein the memory comprises a ternary content-addressable memory (TCAM).
17. The non-transitory machine readable medium of claim 13 , the steps further comprising:
removing the temporary entry from the memory after a period of time;
wherein the period of time is determined based upon a length of time to update the network fabric with the second plurality of addresses.