GENERATING MONOTONE HASH PREFERENCES
Selecting a resource to fulfill a resource requirement is disclosed. For each resource requirement, a resource-specific affinity value is computed with respect to each of a plurality of resources. A bias is applied to each of at least a subset of the resource-specific affinity values. The biased, as applicable, resource-specific affinity values are sorted into a resource preference list. The sorted preference list is used to select a resource to fulfill the resource requirement.
1 . A resource selection system, comprising:
a communication interface configured to receive a communication comprising an operation associated with a data object; and
a processor coupled to the communication interface and configured to:
compute for each of a plurality of resources, with respect to the data object, a resource-specific affinity value;
apply a bias to each of at least a subset of the resource-specific affinity values, wherein the bias comprises a function that is monotonically increasing; and
use the resource-specific affinity values to select a resource to fulfill the operation associated with the data object.
2 . The system of claim 1 , wherein the function is monotonically increasing in terms of one or more of an input hash and a desired bias.
3 . The system of claim 1 , wherein the bias comprises a resource-specific bias.
4 . The system of claim 1 , wherein the bias comprises a resource-specific bias determined for a particular resource based at least in part on a data associated with that particular resource.
5 . The system of claim 1 , wherein the bias is determined based at least in part on a data associated with the operation.
6 . The system of claim 1 , wherein the bias is determined for a particular resource based at least in part on a data reflecting a relationship between the data object and the particular resource.
7 . The system of claim 6 , wherein the bias is determined based at least in part on a stored data reflecting a past relationship between the resource and the data object.
8 . The system of claim 6 , wherein the bias is determined based at least in part on a stored data reflecting a past relationship between the resource and a user with which the operation is associated.
9 . The system of claim 1 , wherein the bias comprises a multiplicative bias.
10 . The system of claim 1 , wherein the operation is associated with storing the data object.
11 . The system of claim 1 , wherein the operation is associated with finding the data object in storage.
12 . The system of claim 1 , wherein the plurality of resources comprises a plurality of storage nodes.
13 . The system of claim 1 , wherein a consistent hash function is used to compute the resource-specific affinity values.
14 . The system of claim 1 , wherein the processor is further configured to:
detect that a removed resource has been removed from the plurality of resources; and
remove a resource-specific affinity value associated with the removed resource from the resource-specific affinity values.
15 . The system of claim 1 , wherein the processor is further configured to:
detect that an added resource has been added to the plurality of resources; and
compute a resource-specific affinity value for the added resource with respect to the data object.
16 . A method of selecting a resource, comprising:
computing, using a processor, for each of a plurality of resources, with respect to the data object, a resource-specific affinity value;
applying a bias to each of at least a subset of the resource-specific affinity values, wherein the bias comprises a function that is monotonically increasing; and
using the resource-specific affinity values to select a resource to fulfill the operation associated with the data object.
17 . The method of claim 16 , wherein a consistent hash function is used to compute the resource-specific affinity values.
18 . The method of claim 16 , further comprising:
detecting that a removed resource has been removed from the plurality of resources; and
removing a resource-specific affinity value associated with the removed resource from the resource-specific affinity values.
19 . The method of claim 16 , further comprising:
detecting that an added resource has been added to the plurality of resources; and
computing a resource-specific affinity value for the added resource with respect to the data object.
20 . A computer program product for selecting a resource, the computer program product being embodied in a computer readable storage medium and comprising computer instructions for:
computing for each of a plurality of resources, with respect to the data object, a resource-specific affinity value;
applying a bias to each of at least a subset of the resource-specific affinity values, wherein the bias comprises a function that is monotonically increasing; and
using the resource-specific affinity values to select a resource to fulfill the operation associated with the data object.