# US 2026/0086714 A1 · System and method for optimizing memory utilization in a distributed file system

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

- Number: US 2026/0086714 A1
- Document ID: US20260086714A1
- Status: Published application
- Kind: A1
- Application number: 18/897,755
- Filed: 2024-09-26
- Published: 2026-03-26
- Earliest claimed: 2024-09-26
- Assignees: Visa International Service Association
- Applicants: Visa International Service Association
- Inventors: Satyam Raj; Shivam Mohan
- Web page: https://patentgrep.com/patent/US20260086714A1

## CPC

- G06F 3/0608: {Saving storage space on storage systems}
- G06F 3/064: {Management of blocks}
- G06F 3/067: {Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]}
- G06F 3/0673: {Single storage device}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method for optimizing memory utilization in a distributed file system is disclosed. The method includes receiving a write request to store a data chunk in at least one node of a plurality of nodes and monitoring memory usage of a data set across the plurality of nodes in the distributed file system. The data set is partitioned into a plurality of data blocks across the plurality of nodes. The method includes determining a current state of data block replications across the plurality of nodes, selecting a target node to store the data chunk, and dynamically adjusting selection of the target node based on historical write request patterns according to a reinforcement learning model.

## Claims

### Claim 1

A method for optimizing memory utilization in a distributed file system, the method comprising:
receiving a write request to store a data chunk in at least one node of a plurality of nodes;
monitoring memory usage of a data set across the plurality of nodes in the distributed file system, wherein the data set is partitioned into a plurality of data blocks across the plurality of nodes;
determining a current state of data block replications across the plurality of nodes;
selecting a target node to store the data chunk; and
dynamically adjusting selection of the target node based on historical write request patterns according to a reinforcement learning model.

### Claim 2 (depends on 1)

The method of claim 1, comprising migrating data blocks from nodes with memory usage exceeding a predefined threshold to nodes with available memory below the predefined threshold.

### Claim 3 (depends on 2)

The method of claim 2, comprising rebalancing a distribution of data blocks to provide optimal memory utilization across the plurality of nodes.

### Claim 4 (depends on 1)

The method of claim 1, wherein dynamically adjusting a selection of a target node comprises selecting a node with minimal memory space that is sufficient to accommodate a data block and minimize wasted memory space.

### Claim 5 (depends on 1)

The method of claim 1, wherein dynamically adjusting a selection of a target node comprises selecting a node with maximum memory space to provide a larger remainder of unused memory space for future use.

### Claim 6 (depends on 1)

The method of claim 1, wherein dynamically adjusting a selection of a target node comprises selecting a first node that has sufficient memory space to accommodate a data block and minimizing time taken to find a suitable node to store the data chunk.

### Claim 7 (depends on 1)

The method of claim 1, comprising determining an optimal set of target nodes for the block replacement operation based on a best fit algorithm and selecting a source node with minimal space sufficient to accommodate a data block to minimize wasted space.

### Claim 8 (depends on 1)

The method of claim 1, comprising determining an optimal set of target nodes for the block replacement operation based on a worst fit algorithm and selecting a source node with maximum space to leave a larger remainder for future use.

### Claim 9 (depends on 1)

The method of claim 1, comprising determining an optimal set of target nodes for the block replacement operation based on a first fit algorithm and selecting a first node that has sufficient space to accommodate a data block to minimize time taken to find a suitable node.

### Claim 10

A system for optimizing memory utilization in a distributed file system, the system comprising:
a control node; and
an intelligent coordination unit (ICU) to communicate with the control node, the ICU comprising:
a source node selector to select a source node based on a write request from the control node, the write request comprising a write request in a contiguous storage space of size R chunked into one or more than one C sized unit of contiguous storage space;
a target node selector to determine an optimal set of target nodes based on the selected source node and to free up storage space in a storage block/unit where replacement will occur; and
a block replacement executor to replace storage blocks/units based on the source node, the target node, and blocks as inputs and to free up required storage space in the selected source node to persist the C sized unit of contiguous storage space.

### Claim 11 (depends on 10)

The system of claim 10, wherein the source node selector is configured to select an optimal source node for a write operation to minimize the number of block replacement operations block/unit replacement executor.

### Claim 12 (depends on 10)

The system of claim 10, wherein the source node selector is configured to dynamically adapt to less optimal source nodes choices for certain write requests based on a reinforcement learning model configured to learn based on a previous experience and historical write request patterns.

### Claim 13 (depends on 10)

The system of claim 10, wherein the target node selector is configured to determine an optimal set of target nodes for the block replacement operation based on a best fit algorithm to select a source node with minimal space sufficient to accommodate a data block to minimize wasted space.

### Claim 14 (depends on 10)

The system of claim 10, wherein the target node selector is configured to determine an optimal set of target nodes for the block replacement operation based on a worst fit algorithm to select a source node with maximum space to leave a larger remainder for future use.

### Claim 15 (depends on 10)

The system of claim 10, wherein the target node selector is configured to determine an optimal set of target nodes for the block replacement operation based on a first fit algorithm to select first node that has sufficient space to accommodate a data block to minimize time taken to find a suitable node.

### Claim 16 (depends on 10)

The system of claim 10, wherein the target node selector is configured to dynamically adjust a selection strategy based on historical write request patterns using a reinforcement learning model to choose a selection algorithm and adapt to changing system usage patterns and optimize for memory utilization.

### Claim 17

A system for optimizing memory utilization in a distributed file system comprising a processor and a non-transitory computer-readable medium coupled to the processor, the non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to:
receive a write request to store a data chunk in at least one node of a plurality of nodes;
monitor memory usage of a data set across the plurality of nodes in the distributed file system, wherein the data set is partitioned into a plurality of data blocks across the plurality of nodes;
determine a current state of data block replications across the plurality of nodes;
select a target node to store the data chunk; and
dynamically adjust selection of the target node based on historical write request patterns according to a reinforcement learning model.

### Claim 18 (depends on 17)

The system of claim 17, wherein the non-transitory computer-readable medium stores additional instructions that, when executed by the processor, cause the processor to migrate data blocks from nodes with memory usage exceeding a predefined threshold to nodes with available memory below the predefined threshold.

### Claim 19 (depends on 17)

The system of claim 17, wherein the non-transitory computer-readable medium stores additional instructions that, when executed by the processor, cause the processor to rebalance a distribution of data blocks to provide optimal memory utilization across the plurality of nodes.

### Claim 20 (depends on 17)

The system of claim 17, wherein the non-transitory computer-readable medium stores additional instructions that, when executed by the processor, cause the processor to:
determine an optimal set of target nodes for the block replacement operation based on a best fit algorithm and selecting a source node with minimal space sufficient to accommodate a data block to minimize wasted space;
determine an optimal set of target nodes for the block replacement operation based on a worst fit algorithm and selecting a source node with maximum space to leave a larger remainder for future use; or
determine an optimal set of target nodes for the block replacement operation based on a first fit algorithm and selecting a first node that has sufficient space to accommodate a data block to minimize time taken to find a suitable node.
