Multi-chassis fabric-backplane enterprise servers
Multi-chassis fabric-backplane enterprise servers include a plurality of chassis managed collectively to form one or more provisioned servers. A central client coordinates gathering of provisioning and management information from the chassis, and arranges for distribution of control information to the chassis. One of the chassis may perform as a host or proxy with respect to information and control communication between the client and the chassis. Server provisioning and management information and commands move throughout the chassis via an Open Shortest Path First (OSPF) protocol. Alternatively, the client may establish individual communication with a subset of the chassis, and directly communicate with chassis in the subset. Server provisioning and management information includes events generated when module status changes, such as when a module is inserted and becomes available, and when a module fails and is no longer available. Each chassis includes a switch fabric enabling communication between chassis modules.
1. A method including the steps of:
communicating multi-chassis control instructions from a multi-chassis control agent to a portal multi-chassis platform process;
distributing the multi-chassis control instructions from the portal multi-chassis platform process to a plurality of peer multi-chassis platform processes executing in a respective plurality of chassis;
coalescing multi-chassis control data from the peer multi-chassis platform processes to the portal multi-chassis platform process;
communicating the multi-chassis control data from the portal multi-chassis platform process to the multi-chassis control agent;
wherein the multi-chassis control agent determines at least a portion of the multi-chassis control instructions based in part on at least a portion of the multi-chassis control data; and
wherein at least one of the chassis includes at least a first partition of a plurality of partitionable symmetric multiprocessors, the first partition being formed at least in part by programmatically configuring links between processors of the physically partitionable symmetric multiprocessor to selectively isolate subsets of the processors with respect to respective shared memory coherency and cache coherency transactions on a link-by-link basis, the multi-chassis control instructions include at least one instruction directed to the first partition, and the multi-chassis control data includes data from the first partition.
2. The method of claim 1 , wherein the distributing of the multi-chassis control instructions is at least in part via an open shortest path first protocol.
3. The method of claim 1 , wherein the coalescing of the multi-chassis control data is at least in part via an open shortest path first protocol.
4. The method of claim 1 , wherein each of the chassis includes a respective switch fabric and at least a portion of the multi-chassis control instructions are communicated in part via at least one of the switch fabrics.
5. The method of claim 1 , wherein each of the chassis includes a respective switch fabric and at least a portion of the multi-chassis control data is communicated in part via at least one of the switch fabrics.
6. The method of claim 1 , further including establishing logical communication between the multi-chassis control agent and the portal multi-chassis platform process.
7. The method of claim 1 , wherein the multi-chassis control agent executes a multi-chassis control process at least in part to determine the portion of the multi-chassis control instructions.
8. The method of claim 1 , wherein the multi-chassis control agent is a client.
9. The method of claim 1 , wherein the portal multi-chassis platform process is a host multi-chassis platform process.
10. The method of claim 1 , wherein at least one of the peer multi-chassis platform processes is a chassis manager process.
11. The method of claim 1 , wherein at least one of the peer multi-chassis platform processes is an enterprise manager process.