초록 ▼

The present invention generally provides a method for grid storage including balancing read and write requests from applications across a first group of nodes in a grid storage system for avoiding hot spots and optimizing performance through smart caching; balancing storage capacity across a second

The present invention generally provides a method for grid storage including balancing read and write requests from applications across a first group of nodes in a grid storage system for avoiding hot spots and optimizing performance through smart caching; balancing storage capacity across a second group of nodes in the grid storage system, nodes in the first and second groups being at least one of hardware interchangeable online, capable of being added to change performance or capacity of the grid storage system and capable of being removed to change performance or capacity of the grid storage system; and self managing of the first and second groups of nodes for providing at least one of scalability, self healing after failure of components in the grid storage, non-disruptive upgrades to the grid storage system, and eliminating duplicate data on an object or sub-object level in the grid storage system.

대표청구항 ▼

1. A method for grid storage comprising the steps of: balancing read and write requests from applications across a first group of nodes in a grid storage system for avoiding hot spots and optimizing performance through smart caching;balancing storage capacity across a second group of nodes in the gr

1. A method for grid storage comprising the steps of: balancing read and write requests from applications across a first group of nodes in a grid storage system for avoiding hot spots and optimizing performance through smart caching;balancing storage capacity across a second group of nodes in the grid storage system, nodes in the first and second groups being at least one of hardware interchangeable online, capable of being added to change performance or capacity of the grid storage system, and capable of being removed to change performance or capacity of the grid storage system; andself managing of the first and second groups of nodes for providing at least one of scalability, self healing after failure of components in the grid storage, non-disruptive upgrades to the grid storage system, and eliminating duplicate data on an object or sub-object level in the grid storage system;wherein the balancing steps with the first and second group of nodes comprises generating a plurality of data fragments by erasure coding a block of data and storing each of the plurality of data fragments in different physical nodes so they can be moved to maximize utilization of storage capacity; andwherein generating a plurality of data fragments by erasure coding a block of data comprises:determining a network cardinality;dividing the block of data into a number of original fragments; andcreating a plurality of redundant fragments wherein the number of redundant fragments is equal to the network cardinality minus the number of original data fragments. 2. The method of claim 1, wherein the first and second groups of nodes and the self managing are configured without a central database or centralized lookup table for ensuring no single point of failure in the grid storage system while enabling scalability. 3. The method of claim 1, wherein storing each of the plurality of data fragments in different physical nodes comprises: logically dividing each of the physical nodes into a plurality of slots; andstoring each of the plurality of data fragments in different slots on different physical nodes. 4. The method of claim 1, further comprising associating the different slots on different physical nodes as a virtual node. 5. The method of claim 1, further comprising mapping storage locations of the plurality of data fragments in a data structure wherein the storage locations are the physical nodes in which the plurality of data fragments are stored. 6. The method of claim 5, wherein the data structure is a distributed hash table. 7. The method of claim 1, wherein the self managing comprises content and metadata of any piece of data written to the grid storage system being indexed so the content and metadata can be searched and classified into classifications for allowing automatic application of user policies. 8. The method of claim 1, wherein the self managing comprises maintaining rules that govern the life cycle of each piece of data with attributes assignable based on metadata or by an automated classification based on content, the attributes establishing importance of the data for affecting level of protection of the data by the grid storage system. 9. The method of claim 1, wherein the self managing comprises data on the storage nodes being split into chunks on a sub-file level and a hash process calculating a key serving as an identifier for a specific chunk of data. 10. The method of claim 9, wherein if chunks are already in the grid system because a version of the same or similar file has been stored previously, the chunk is discarded, with new chunks getting stored for ensuring one instance of each chunk taking up storage space on the storage nodes. 11. The method of claim 1, wherein the self managing comprises data on the storage nodes being split into chunks on a sub-file level, breaking each chunk into a set number of original fragments, for each of the fragments calculating a number of parity fragments equal to a desired level of protection from data loss. 12. The method of claim 11, wherein level of protection from data loss can be modified by modifying number of parity fragments. 13. The method of claim 11, wherein once parity fragments are created they are written to storage in a way to spread the fragments across different storage nodes and disks within a storage node for ensuring that disk or storage node failures result in only one fragment lost per chunk.