Read optimization operations in a storage system

• US 9,864,761 B1
• Filed: 08/08/2014
• Issued: 01/09/2018
• Est. Priority Date: 08/08/2014
• Status: Active Grant

• Current Assignee: Pure Storage, Inc.
• Original Assignee: Pure Storage, Inc.
• Inventors: Golden, Christopher, Grunwald, David, Cao, Jianting
• Primary Examiner(s): Pham, Tuan A
• Application Number:
  US14/455,121
• Time in Patent Office: 1,250 Days
• Field of Search: None
• CPC Class Codes: G06F 16/21 Design, administration or maintenance of databases
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/064 Management of blocks
• CPC Class Codes: G06F 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD

Abstract

A system and method for performing read optimization of a volume while allowing user operations to target the volume. Read optimization is prevented from being performed for a top level of the medium graph for a given volume, wherein the top level is in a read-write state. Rather than waiting for the given volume to be idle so as to perform read optimization, read optimization is run at lower levels in the medium graph of the given volume. This allows user operations to modify the medium graph of the top level of the given volume while simultaneously read optimization is being run on mediums which underlie the top level.

Claims

1. A system comprising:

one or more storage devices; and

a storage controller coupled to the one or more storage devices, wherein responsive to selecting a dataset to be read optimized, the storage controller is configured to;

identify metadata associated with the dataset, wherein the metadata includes;

an active logical grouping of data that is in a read-write state and corresponds to an anchor logical grouping of data; and

non-active logical groupings of data that are in a read-only state and correspond to one or more logical data groupings which underlie the anchor logical grouping of data;

based on the metadata associated with the dataset, select one or more of the non-active logical groupings of data for a read optimization operation; and

perform a user operation on the dataset using the active logical grouping of data that is in the read-write state, while simultaneously;

performing the read optimization operation on the selected one or more non-active logical groupings of data by reducing the number of mappings within the non-active logical groupings of data; and

preventing the read optimization operation from being performed on the active logical grouping of data in the read-write state.

2. The system as recited in claim 1, wherein reducing the number of mappings within the non-active logical groupings of data comprises collapsing a particular medium to make the particular medium point to itself.

3. The system as recited in claim 1, wherein reducing the number of mappings within the non-active logical groupings of data comprises modifying address translation table entries for a particular medium such that blocks corresponding to the particular medium are mapped directly from the particular medium.

4. The system as recited in claim 1, wherein the storage controller is configured to select the dataset to be read optimized responsive to at least one of:

an amount of metadata has reached a given threshold, read access latency is longer than desired, a scheduled time has arrived, a predetermined period of time since a last read optimization has elapsed, and detecting a number of mediums has reached threshold.

5. A method comprising:

selecting a dataset to be read optimized;

responsive to selecting the dataset to be read optimized;

identifying metadata associated with the dataset, wherein the metadata includes;

an active logical grouping of data that is in a read-write state and corresponds to an anchor logical grouping of data; and

non-active logical groupings of data that are in a read-only state and correspond to one or more logical data groupings which underlie the anchor logical grouping of data;

based on the metadata associated with the dataset, selecting one or more of the non-active logical groupings of data for a read optimization operation; and

performing a user operation on the dataset using the active logical grouping of data that is in the read-write state, while simultaneously;

performing the read optimization operation on the selected one or more non-active logical groupings of data by reducing the number of mappings within the non-active logical groupings of data; and

preventing the read optimization operation from being performed on the active logical grouping of data in the read-write state.

6. The method as recited in claim 5, wherein reducing the number of mappings within the non-active logical groupings of data comprises collapsing a particular medium to make the particular medium point to itself.

7. The method as recited in claim 5, wherein reducing the number of mappings within the non-active logical groupings of data comprises consolidating address translation table entries for a particular medium, resulting in all blocks of the particular medium being mapped directly from the particular medium.

8. The method as recited in claim 5, further comprising selecting the dataset to be read optimized responsive to at least one of:

an amount of metadata has reached a given threshold, read access latency is longer than desired, a scheduled time has arrived, a predetermined period of time since a last read optimization has elapsed, and detecting a number of mediums has reached threshold.

9. A non-transitory computer readable storage medium storing program instructions, wherein the program instructions are executable by a processor to:

select a dataset to be read optimized;

responsive to selecting the dataset to be read optimized;

identify metadata associated with the dataset, wherein the metadata includes;

an active logical grouping of data that is in a read-write state and corresponds to an anchor logical grouping of data; and

non-active logical groupings of data that are in a read-only state and correspond to one or more logical data groupings which underlie the anchor logical grouping of data;

based on the metadata associated with the dataset, select one or more of the non-active logical groupings of data for a read optimization operation; and

perform a user operation on the dataset using the active logical grouping of data that is in the read-write state, while simultaneously;

performing the read optimization operation on the selected one or more non-active logical groupings of data by reducing the number of mappings within the non-active logical groupings of data; and

preventing the read optimization operation from being performed on the active logical grouping of data in the read-write state.

10. The non-transitory computer readable storage medium as recited in claim 9, wherein reducing the number of mappings within the non-active logical groupings of data comprises collapsing a particular medium to make the particular medium point to itself.

11. The non-transitory computer readable storage medium as recited in claim 9, wherein reducing the number of mappings within the non-active logical groupings of data comprises consolidating address translation table entries for particular medium, resulting in all blocks of the particular medium being mapped directly from the particular medium.