System and method for state preservation in a stretch cluster
AssigneeDell Products L.P.
InventorsVictor V. Mashayekhi, Tau Leng
Life of this patent
Estimating the term…
Abstract
A system and method for preserving the data content of a network having a cluster is provided in which each of the network interface controllers of the servers of the cluster includes a copy of a set of data. Each of the network interface controllers includes a local power source, local memory, and an operating system, allowing the network interface controllers to remain in communication despite a failure to one or both of the associated servers. The data may be operational data that is critical to the operation of the cluster system. Alterations to the set of data can only be made by a write operation that is logically atomic or indivisible. Incomplete write operations are discarded unless a successful write operation is performed to both servers.
Claims11
A method for updating the content of a set of data accessible by a server of a cluster system, comprising the steps of:
- providing first and second servers;
- providing a network controller associated with each server, each network controller including a local power source and a local memory, wherein the local power source comprises a standby power source operable to provide power to local memory and a processor of the network controller in the event of a power failure in the network controller, and wherein the network controllers associated with each of the first and the second servers are operable to continue to communicate with each other despite an operational failure in at least one of the first server and the second server;
- storing in the local memory of each network controller a copy of a set of data;
- and
- performing a write operation to the memory of each network controller such that the content of the set of data of each network controller is the same.
A cluster system comprising:
- a first server including a network controller;
- a second server including a network controller;
- wherein the network controllers associated with each of the first and the second servers continue to communicate with each other despite an operational failure in at least one of the first server and the second server;
- wherein the network controller of the first and second servers each include a copy of a set of data describing the status of the cluster system;
- and
- wherein the network controllers of the first server and the second server are each coupled to a standby power source operable to provide standby power to the network controller such that the set of data is accessible despite the loss of operational power in the associated server and wherein the network controllers associated with each of the first and the second servers are operable to communicate with each other despite an operational failure in at least one of the first server and the second server.
Dependencies are inferred from the claim text.