Back to results
US 8,554,993 B2Granted
Distributed content storage and retrieval
AssigneeTranslattice, Inc.
InventorsRobert F. Ross, Michael P. Lyle
Life of this patent
Estimating the term…
Abstract
Distributed content storage and retrieval is disclosed. A set of features associated with a content object is determined. A storage location is selected to perform an operation with respect to the content object, from a plurality of storage locations comprising a distributed content storage system, based at least in part on probability data indicating a degree to which the selected storage location is associated statistically with a feature comprising the set of features determined to be associated with the content object.
Claims47
A method of storing data, comprising:
- determining a set of features associated with a content object;
- selecting a storage location to perform an operation with respect to the content object, from a plurality of storage locations within a distributed content storage system, based at least in part on probability data indicating a degree to which the selected storage location is associated statistically with a feature within the set of features determined to be associated with the content object, wherein the probability data is determined at least in part by gathering at each of the plurality of storage locations an access count data reflecting for each feature within a plurality of features how many objects having that feature have been accessed from that location, wherein changes in access statistics are damped by reducing a degree to which such changes are reflected in the access count data, wherein the changes in access statistics are damped by estimating or determining the proportion of objects that will no longer be in their preferred locations with a given new access profile;
- and
- applying an optimization algorithm configured to balance convergence towards the new access profile against the proportion of objects that will no longer be in their preferred locations.
1 more element
A distributed content storage system, comprising:
- a storage device configured to store content objects;
- and
- processor coupled to the storage device and configured to:
- determine a set of features associated with a content object;
- select a storage location at which to perform an operation with respect to the content object, from a plurality of storage locations within a distributed content storage system, based at least in part on probability data indicating a degree to which the selected storage location is associated statistically with a feature within the set of features determined to be associated with the content object, wherein the probability data is determined at least in part by gathering at each of the plurality of storage locations an access count data reflecting for each feature within a plurality of features how many objects having that feature have been accessed from that location, wherein changes in access statistics are damped by reducing a degree to which such changes are reflected in the access count data, wherein the changes in access statistics are damped by estimating or determining the proportion of objects that will no longer be in their preferred locations with a given new access profile;
- and
- apply an optimization algorithm configured to balance convergence towards the new access profile against the proportion of objects that will no longer be in their preferred locations.
4 more elements
A computer program product for storing content objects, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
- determining a set of features associated with a content object;
- selecting a storage location to perform an operation with respect to the content object, from a plurality of storage locations within a distributed content storage system, based at least in part on probability data indicating a degree to which the selected storage location is associated statistically with a feature within the set of features determined to be associated with the content object, wherein the probability data is determined at least in part by gathering at each of the plurality of storage locations an access count data reflecting for each feature within a plurality of features how many objects having that feature have been accessed from that location, wherein changes in access statistics are damped by reducing a degree to which such changes are reflected in the access count data, wherein the changes in access statistics are damped by estimating or determining the proportion of objects that will no longer be in their preferred locations with a given new access profile;
- and
- applying an optimization algorithm configured to balance convergence towards the new access profile against the proportion of objects that will no longer be in their preferred locations.
1 more element
Dependencies are inferred from the claim text.