Selecting data nodes for WAN caching in a hybrid cloud environment
AssigneeInternational Business Machines Corporation
InventorsAbhishek Jain, Sasikanth Eda, Sandeep Ramesh Patil, Anil Laxman Palled
Life of this patent
Estimating the term…
Abstract
A computer-implemented method for storing blocks of a file in a heterogeneous environment. The method may comprise receiving a command header for an operation associated with a file from a first data node, matching the enhanced metadata with capabilities of a second data node, and generating a block layout map for the file based at least in part on the enhanced metadata and the capabilities of the second data node. The command header may include enhanced metadata associated with a file, and the enhanced metadata may comprise a type of workload that generated the file and a data access pattern.
Claims20
A method for storing blocks of a file in a heterogeneous environment, comprising:
- receiving a command header for an operation associated with a file from a first data node, wherein the command header includes enhanced metadata associated with a file, wherein the enhanced metadata comprises a type of workload that generated the file and a data access pattern;
- matching the enhanced metadata with capabilities of a second data node;
- and
- generating a block layout map for the file based at least in part on the enhanced metadata and the capabilities of the second data node.
A computer-implemented method for automatically selecting a data node in a distributed storage network based on data replication request, comprising;
- identifying a current hardware configuration of a data node, the hardware configuration including a memory type, a memory amount, a storage type, and available coprocessors;
- dynamically updating a metadata nodes with the current hardware configuration;
- receiving credentials for a remote data node to read a local file, wherein the remote data node is selected based at least in part a workload hardware configuration requirement associated with the file;
- and
- writing the file to the remote data node.
A distributed cluster file system, comprising:
- a first plurality of data nodes associated with a local cloud;
- a second plurality of data nodes associated with a remote cloud;
- and
- a metadata server adapted to:
- receive a command header for an operation associated with a file from one of the first plurality of data nodes, wherein the command header includes enhanced metadata associated with a file, wherein the enhanced metadata comprises a type of workload that generated the file and a data access pattern;
- match the enhanced metadata with capabilities of one or more of the second plurality of data nodes;
- and
- generate a block layout map for the file based at least in part on the enhanced metadata and the capabilities of the second data node.
Dependencies are inferred from the claim text.