# US 2020/0004623 A1 · Determining when to perform error checking of a storage unit by using a machine learning module

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0004623 A1
- Document ID: US20200004623A1
- Status: Published application
- Kind: A1
- Application number: 16/023,443
- Filed: 2018-06-29
- Published: 2020-01-02
- Earliest claimed: 2018-06-29
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Matthew G. Borlick; Karl A. Nielsen; Clint A. Hardy; Lokesh M. Gupta
- Web page: https://patentgrep.com/patent/US20200004623A1

## CPC

- G06F 11/106: {Correcting systematically all correctable errors, i.e. scrubbing}
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06F 11/1096: {Parity calculation or recalculation after configuration or reconfiguration of the system}
- G06N 20/00: Machine learning
- G06F 11/0751: {Error or fault detection not based on redundancy (power supply failures G06F1/30; network fault management H04L41/06)}
- G06F 11/0754: {by exceeding limits}
- G06F 11/0763: {by bit configuration check, e.g. of formats or tags}
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 5/025: Extracting rules from data
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Provided are a computer program product, system, and method for using a machine learning module to determine when to perform error checking of a storage unit. Input on attributes of at least one storage device comprising the storage unit are provided to a machine learning module to produce an output value. An error check frequency is determined from the output value. A determination is made as to whether the error check frequency indicates to perform an error checking operation with respect to the storage unit. The error checking operation is performed in response to determining that the error checking frequency indicates to perform the error checking operation.

## Claims

### Claim 1

A computer program product for error checking data in a storage unit, the computer program product comprising a computer readable storage medium storing computer readable program code that when executed performs operations, the operations comprising:
providing input on attributes of at least one storage device comprising the storage unit to a machine learning module to produce an output value;
determining an error check frequency from the output value;
determining whether the error check frequency indicates to perform an error checking operation with respect to the storage unit; and
performing the error checking operation in response to determining that the error checking frequency indicates to perform the error checking operation.

### Claim 2 (depends on 1)

The computer program product of claim 1, wherein the storage unit comprises one of an array of storage devices or one storage device.

### Claim 3 (depends on 1)

The computer program product of claim 1, wherein the providing the input on attributes of the storage unit comprises periodically providing the input to the machine learning module to periodically produce the output value, wherein the operations of determining the error check frequency and the determining whether the error check frequency indicates to perform the error checking operation.

### Claim 4 (depends on 1)

The computer program product of claim 1, wherein the determining the error check frequency from the output value comprises determining a number of writes as a function of the output value, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
determining whether a write counter exceeds the error check frequency, wherein the error checking operation is performed in response to the write counter exceeding the error check frequency.

### Claim 5 (depends on 4)

The computer program product of claim 4, wherein the operations further comprise:
resetting the write counter to zero in response to determining to perform the error checking operation.

### Claim 6 (depends on 1)

The computer program product of claim 1, wherein the output value comprises a number from zero to 1 indicating a likelihood that there is an error in the storage unit, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
not performing the error checking operation in response to the output value being less than a lower bound;
performing the error checking operation in response to the output value being greater than an upper bound; and
adjusting the error check frequency based on the output value in response to the output value being between the lower bound and the upper bound.

### Claim 7 (depends on 6)

The computer program product of claim 6, wherein the error checking operation checks a fixed number of last writes, and wherein the adjusting the error check frequency sets the error check frequency to the fixed number of last writes divided by the output value.

### Claim 8 (depends on 1)

The computer program product of claim 1, wherein the attributes of the at least one storage used as the input to the machine learning module include at least one of:
an error type if the error checking operation detected an error during a last run of the error checking operation or indication of no error if the error checking operation did not detect an error during the last run;
a type of at least one storage device comprising the storage unit;
an age of the at least one storage device from first use;
a firmware level of the at least one storage device;
a read operations per second at the at least one storage device; and
a write operations per second at the at least one storage device.

### Claim 9 (depends on 1)

The computer program product of claim 1, wherein the operations further comprise:
incrementing a write counter in response to performing a write operation against the storage unit;
determining whether the write counter satisfies a condition with respect to the error checking frequency; and
performing the error checking operation and resetting the write counter in response to determining that the write counter satisfies the condition with respect to the error checking frequency.

### Claim at position 10

10-15. (canceled)

### Claim 16

A system computer for error checking data in a storage unit, comprising:
a processor; and
a computer readable storage medium storing computer readable program code that when executed performs operations, the operations comprising:
providing input on attributes of at least one storage device comprising the storage unit to a machine learning module to produce an output value;
determining an error check frequency from the output value;
determining whether the error check frequency indicates to perform an error checking operation with respect to the storage unit; and
performing the error checking operation in response to determining that the error checking frequency indicates to perform the error checking operation.

### Claim 17 (depends on 16)

The system of claim 16, wherein the determining the error check frequency from the output value comprises determining a number of writes as a function of the output value, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
determining whether a write counter exceeds the error check frequency, wherein the error checking operation is performed in response to the write counter exceeding the error check frequency.

### Claim 18 (depends on 16)

The system of claim 16, wherein the output value comprises a number from zero to 1 indicating a likelihood that there is an error in the storage unit, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
not performing the error checking operation in response to the output value being less than a lower bound;
performing the error checking operation in response to the output value being greater than an upper bound; and
adjusting the error check frequency based on the output value in response to the output value being between the lower bound and the upper bound.

### Claim 19 (depends on 16)

The system of claim 16, wherein the operations further comprise:
incrementing a write counter in response to performing a write operation against the storage unit;
determining whether the write counter satisfies a condition with respect to the error checking frequency; and
performing the error checking operation and resetting the write counter in response to determining that the write counter satisfies the condition with respect to the error checking frequency.

### Claim at position 15

20-22. (canceled)

### Claim 23

A method for error checking data in a storage unit, comprising:
providing input on attributes of at least one storage device comprising the storage unit to a machine learning module to produce an output value;
determining an error check frequency from the output value;
determining whether the error check frequency indicates to perform an error checking operation with respect to the storage unit; and
performing the error checking operation in response to determining that the error checking frequency indicates to perform the error checking operation.

### Claim 24 (depends on 23)

The method of claim 23, wherein the determining the error check frequency from the output value comprises determining a number of writes as a function of the output value, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
determining whether a write counter exceeds the error check frequency, wherein the error checking operation is performed in response to the write counter exceeding the error check frequency.

### Claim 25 (depends on 23)

The method of claim 23, wherein the output value comprises a number from zero to 1 indicating a likelihood that there is an error in the storage unit, wherein the determining whether the error check frequency indicates to perform the error checking operation comprises:
not performing the error checking operation in response to the output value being less than a lower bound;
performing the error checking operation in response to the output value being greater than an upper bound; and
adjusting the error check frequency based on the output value in response to the output value being between the lower bound and the upper bound.

### Claim 26 (depends on 23)

The method of claim 23, further comprising:
incrementing a write counter in response to performing a write operation against the storage unit;
determining whether the write counter satisfies a condition with respect to the error checking frequency; and
performing the error checking operation and resetting the write counter in response to determining that the write counter satisfies the condition with respect to the error checking frequency.

### Claim 27 (depends on 16)

The system of claim 16, wherein the providing the input on attributes of the storage unit comprises periodically providing the input to the machine learning module to periodically produce the output value, wherein the operations of determining the error check frequency and the determining whether the error check frequency indicates to perform the error checking operation.

### Claim 28 (depends on 16)

The system of claim 16, wherein the attributes of the at least one storage used as the input to the machine learning module include at least one of:
an error type if the error checking operation detected an error during a last run of the error checking operation or indication of no error if the error checking operation did not detect an error during the last run;
a type of at least one storage device comprising the storage unit;
an age of the at least one storage device from first use;
a firmware level of the at least one storage device;
a read operations per second at the at least one storage device; and
a write operations per second at the at least one storage device.

### Claim 29 (depends on 18)

The system of claim 18, wherein the error checking operation checks a fixed number of last writes, and wherein the adjusting the error check frequency sets the error check frequency to the fixed number of last writes divided by the output value.

### Claim 30 (depends on 23)

The method of claim 23, wherein the attributes of the at least one storage used as the input to the machine learning module include at least one of:
an error type if the error checking operation detected an error during a last run of the error checking operation or indication of no error if the error checking operation did not detect an error during the last run;
a type of at least one storage device comprising the storage unit;
an age of the at least one storage device from first use;
a firmware level of the at least one storage device;
a read operations per second at the at least one storage device; and
a write operations per second at the at least one storage device.

### Claim 31 (depends on 25)

The method of claim 25, wherein the error checking operation checks a fixed number of last writes, and wherein the adjusting the error check frequency sets the error check frequency to the fixed number of last writes divided by the output value.

### Claim 32

A computer program product for error checking data in a storage unit, the computer program product comprising a computer readable storage medium storing computer readable program code that when executed performs operations, the operations comprising:
providing input on attributes of at least one storage device comprising the storage unit to a machine learning module to produce an output value;
determining whether the output value indicates to perform an error checking operation with respect to the storage unit; and
performing the error checking operation in response to determining that the output value indicates to perform the error checking operation.

### Claim 33 (depends on 32)

The computer program product of claim 32, wherein the operations further comprise:
in response to determining that the output value does not indicate to perform the error checking operation, set an error checking frequency based on a number of writes and the output value; and
performing the error checking operation in response to determining that the error checking frequency indicates to perform the error checking operation.

### Claim 34

A system for error checking data in a storage unit, comprising:
a processor; and
a computer readable storage medium storing computer readable program code that when executed performs operations, the operations comprising:
providing input on attributes of at least one storage device comprising the storage unit to a machine learning module to produce an output value;
determining whether the output value indicates to perform an error checking operation with respect to the storage unit; and
performing the error checking operation in response to determining that the output value indicates to perform the error checking operation.

### Claim 35 (depends on 34)

The system of claim 34, wherein the operations further comprise:
in response to determining that the output value does not indicate to perform the error checking operation, set an error checking frequency based on a number of writes and the output value; and
performing the error checking operation in response to determining that the error checking frequency indicates to perform the error checking operation.
