Back to results
US 10,353,616 B1Granted
Managing data relocation in storage systems
AssigneeEMC IP Holding Company LLC
InventorsQin Tao, Jun Zeng, Khang Can, Xiangping Chen
Life of this patent
Estimating the term…
Abstract
A method is used in managing data relocation in storage systems. Data access activity information is gathered for a set of slices of a storage tier in a data storage system for migrating the set of slices from the storage tier to another storage tier. The data storage system includes a first storage tier and a second storage tier configured such that performance characteristics associated with the first storage tier is superior to the second storage tier. Based on a pattern indicated by the data access activity information, a temperature for the set of slices is determined by applying a predictive analysis technique. Based on the determination, relocation of data is effected in the storage system.
Claims16
A method for use in managing data relocation in storage systems, the method comprising:
- gathering a set of data access activity information over a long period of time for a set of slices of a storage tier in a data storage system periodically at a specific time interval for migrating the set of slices from the storage tier to another storage tier, wherein the data storage system includes a first storage tier and a second storage tier configured such that performance characteristics associated with the first storage tier is superior to the second storage tier, wherein the set of data access activity information includes timestamp information for each I/O activity for detecting a time-based I/O pattern;
- determining whether a long term I/O pattern indicating a long term trend exists by evaluating the set of data access activity information, wherein the long term I/O pattern includes the time-based I/O pattern based on a specific pattern of access of data during a specific period of time, wherein the time-based I/O pattern indicates that data of a slice, is accessed repeatedly during the specific period of time over the long period of time indicating the long term trend;
- based on the long term I/O pattern determined from the set of data access activity information, determining a respective predictive temperature and a confidence factor associated with the respective predictive temperature for each slice of the set of slices by applying a predictive analysis technique on the set of data access activity information, wherein the confidence factor for a slice indicates accuracy and an error margin with which the long term I/O pattern is determined for the slice, and is computed based on a set of factors, the set of factors including a number of samples gathered in the set of data access activity information and whether the long term I/O pattern is determined from the set of data access activity information, wherein a respective predictive temperature is computed based on a weight assigned to a temperature based on the confidence factor with which the long term I/O pattern is determined;
- and
- based on the determination, managing relocation of each slice of the set of slices based on the respective predictive temperatures determined for respective slices upon detecting the long term I/O pattern with a high confidence factor, wherein each slice of the set of slices is proactively relocated based on a respective predictive temperature and the confidence factor determined for the respective slice, wherein proactively relocating a slice enables the data storage system to relocate the slice to a hot storage tier prior to an increase in I/O activity for the slice.
2 more elements
A system for use in managing data relocation in storage systems, the system comprising a processor configured to:
- gather a set of data access activity information over a long period of time for a set of slices of a storage tier in a data storage system periodically at a specific time interval for migrating the set of slices from the storage tier to another storage tier, wherein the data storage system includes a first storage tier and a second storage tier configured such that performance characteristics associated with the first storage tier is superior to the second storage tier, wherein the set of data access activity information includes timestamp information for each I/O activity for detecting a time-based I/O pattern;
- determine whether a long term I/O pattern indicating a long term trend exists by evaluating the set of data access activity information, wherein the long term I/O pattern includes the time-based I/O pattern based on a specific pattern of access of data during a specific period of time, wherein the time-based I/O pattern indicates that data of a slice is accessed repeatedly during the specific period of time over the long period of time indicating the long term trend;
- based on the long term I/O pattern determined from the set of data access activity information, determine a respective predictive temperature and a confidence factor associated with the respective predictive temperature for each slice of the set of slices by applying a predictive analysis technique on the set of data access activity information, wherein the confidence factor for a slice indicates accuracy and an error margin with which the long term I/O pattern is determined for the slice, and is computed based on a set of factors, the set of factors including a number of samples gathered in the set of data access activity information and whether the long term I/O pattern is determined from the set of data access activity information, wherein a respective predictive temperature is computed based on a weight assigned to a temperature based on the confidence factor with which the long term I/O pattern is determined;
- and
- based on the determination, manage relocation of each slice of the set of slices based on the respective predictive temperatures determined for respective slices upon detecting the long term I/O pattern with a high confidence factor, wherein each slice of the set of slices is proactively relocated based on a respective predictive temperature and the confidence factor determined for the respective slice, wherein proactively relocating a slice enables the data storage system to relocate the slice to a hot storage tier prior to an increase in I/O activity for the slice.
2 more elements
Dependencies are inferred from the claim text.