Distributed system partition

1. A method for client service in a distributed switch, comprising:

maintaining local and global connection state information between a primary and a secondary controlling fibre channel (FC) over Ethernet (FCoE) Forwarders (FCFs) or FC forwarder in a distributed switch;

detecting a partition in the distributed switch, wherein the partition is a split of a system into at least two sides and nodes in one side are unaware of a control state of nodes in another side; and

continuing service to all subtended clients of the distributed switch using local state information, wherein continuing service to all subtended clients of the distributed switch includes:

providing an alternate address to route frames to an appropriate side of the partition via an appropriate one of the primary or the secondary FCF in the partition; and

providing a tunneling mechanism between the appropriate one of the primary or the secondary controlling FCF and an end switching hop where a destination client resides;

wherein the end switching hop uses the tunneling mechanism, even in absence of the partition, to select one of the primary or the secondary FCF for routing of packets to a destination host to control and balance routing traffic in the distributed system.

2. The method of claim 1 , wherein providing the alternate address and providing the tunneling mechanism comprises continuing service to each side of the partition to prevent a loss of frame traffic while re-establishing a link that caused the partition.

3. The method of claim 1 , wherein:

providing the alternate address includes adding an encapsulation header to a frame received by an FCF in the partition to identify a unique address to the appropriate one of the primary or the secondary controlling FCF; and

providing the tunneling mechanism includes using the unique address in the encapsulation header to tunnel the frame from the end switching hop where the destination client resides.

4. The method of claim 3 , wherein providing the alternate address, comprises:

partitioning an area identifier and a port identifier portion of an address space between a principal domain and an alternate address space of the distributed switch.

5. The method of claim 3 , wherein providing the alternate address comprises:

providing alternate domain IDs within a fabric to the primary and the secondary FCFs in a virtual domain of the distributed switch; and

creating a one to one mapping between the virtual domain and the alternate domain IDs.

6. The method of claim 4 , wherein the method, comprises:

creating a table within the primary and the secondary FCFs to map a virtual domain source ID for an incoming frame incoming to the primary or the secondary FCFs frame with an outer header source ID encapsulated to the incoming frame which uniquely identifies the primary and the secondary FCFs.

7. A system, comprising:

a distributed switch seen as a single entity to a network fabric, the distributed switch having a first controlling fibre channel over Ethernet forwarder (FCF) and a second controlling FCF, the distributed switch having links to the network fabric; and

a processor coupled to memory and logic and configured to:

provide an alternate address and encapsulation to route frames to an appropriate side of a partition of the distributed switch, wherein the partition is a split of a system into at least two sides and nodes in one side are unaware of a control state of nodes in another side;

apply a tunneling mechanism to use the alternate address to tunnel between the appropriate one of the first or the second controlling FCF and an end switching hop where a destination client resides; and

apply the tunneling mechanism such that the end switching hop can use the tunneling mechanism to select one of the primary or the secondary FCFs for routing of packets to a destination host to load balance in the distributed switch and the system.

8. The system of claim 7 , wherein the logic of the first and the second controlling FCFs comprises logic to:

continue service to each side of the partition to prevent a loss of frame traffic when the partition occurs in the distributed switch, while re-establishing a link that caused the partition.

9. The system of claim 7 , wherein the first and the second controlling FCFs include logic to encapsulate a frame received to the first or the second controlling FCF with the alternate address when the partition to the distributed switch is detected.

10. The system of claim 9 , wherein the first and the second controlling FCFs include logic to process a tunneled frame to a destination client when the frame arrives encapsulated with the alternate address.

11. The system of claim 7 wherein the logic of the first and the second controlling FCFs comprises logic to:

provide a tunneling mechanism between the appropriate one of the first or the second controlling FCF and an end switching hop where a destination client resides.

12. The system of claim 7 wherein the logic of the first and the second controlling FCFs comprises logic to:

add an encapsulation header to a frame received by an FCF in the partition to identify a unique address to the appropriate one of the first or the second controlling FCF; and

use the unique address in the encapsulation header to tunnel a frame from the end switching hop.

13. The system of claim 12 wherein the logic of the first and the second controlling FCFs comprises logic to:

partition an area identifier and a port identifier portion of an address space between a principal domain and an alternate address space of the distributed switch.

14. The system of claim 12 wherein the logic of the first and the second controlling FCFs comprises logic to:

provide alternate domain IDs within a fabric to the first and the second FCFs in a virtual domain of the distributed switch; and

create a one to one mapping between the virtual domain and the alternate domain IDs.

15. The system of claim 13 wherein the logic of the first and the second controlling FCFs comprises logic to:

create a table within the first and the second FCFs to map a virtual domain source ID for an incoming frame incoming to the first or the second FCFs frame with an outer header source ID encapsulated to the incoming frame which uniquely identifies the first and the second FCFs.

16. A distributed switch in a network, comprising:

a first fibre channel over Ethernet forwarder (FCF);

a second FCF;

a link connecting the first and the second FCFs to exchange global state information between with the first the second FCFs, and

a processor coupled to memory and logic and configured to:

receive a frame from a port;

detect a partition in the distributed switch wherein the partition is a split of a system into at least two sides and nodes in one side are unaware of a control state of nodes in another side;

maintain a table in each FCF to map a source or a destination address for the distributed switch in the network to a different address for the first or the second controlling FCF and

continue service to all subtended clients of the distributed switch using local state information, wherein continuing service to all subtended clients of the distributed switch includes:

providing an alternate address to route frames to an appropriate side of the partition via an appropriate one of the first or the second FCF in the partition; and

providing a tunneling mechanism between the appropriate one of the first or the second controlling FCF and an end switching hop where a destination client resides;

wherein the end switching hop uses the tunneling mechanism, even in absence of the partition, to select one of the primary or the secondary FCF for routing of packets to a destination host to control and balance routing traffic in the distributed system.

17. The distributed switch of claim 16 , wherein the first and the second controlling FCF include logic to remove an encapsulation header of the frame routed to the first or the second FCF and route the frame to a port where a destination client resides.