# US 2026/0186981 A1 · Cache Replacement Method, Apparatus, and System

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

- Number: US 2026/0186981 A1
- Document ID: US20260186981A1
- Status: Published application
- Kind: A1
- Application number: 19/552,621
- Filed: 2026-02-27
- Published: 2026-07-02
- Earliest claimed: 2023-08-28
- Assignees: Huawei Cloud Computing Technologies Co., Ltd.
- Applicants: Huawei Cloud Computing Technologies Co., Ltd.
- Inventors: Pengfei Zuo
- Web page: https://patentgrep.com/patent/US20260186981A1

## CPC

- G06F 12/0868: Data transfer between cache memory and other subsystems, e.g. storage devices or host systems
- G06F 12/121: using replacement algorithms
- G06F 9/50: Allocation of resources, e.g. of the central processing unit [CPU]
- G06F 12/126: with special data handling, e.g. priority of data or instructions, handling errors or pinning
- G06F 2212/1021: Hit rate improvement
- G06F 2212/1024: Latency reduction
- G06F 2212/163: Server or database system
- G06F 2212/254: Distributed memory
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A cache replacement method includes a compute node in a distributed system that sends a first read operation request to a memory node. Then, the compute node separately calculates popularities of the plurality of data objects based on the access information of the plurality of data objects using a first cache replacement algorithm in a plurality of cache replacement algorithms. Finally, the compute node sends a cache replacement request to the memory node based on the popularities of the plurality of data objects, to indicate to replace cold data in the plurality of data objects with a data object to be inserted into the memory node. In this way, the compute node executes the cache replacement algorithm based on global access information.

## Claims

### Claim 1

A method applied to a compute node in a distributed system of a disaggregated memory architecture, wherein the method comprises:
sending a first read operation request to a memory node of the disaggregated memory architecture, wherein the first read operation request requests reading access information of a plurality of data objects from the memory node, and wherein the memory node stores cache data that comprises the plurality of data objects;
separately calculating popularities of the plurality of data objects based on the access information using a first cache replacement algorithm in a plurality of cache replacement algorithms; and
sending, when cache replacement is to be performed on the cache data, a cache replacement request to the memory node based on the popularities,
wherein the cache replacement request instructs to replace cold data in the plurality of data objects with a data object to be inserted into the memory node.

### Claim 2 (depends on 1)

The method of claim 1, wherein the access information comprises a last access timestamp and a quantity of access times.

### Claim 3 (depends on 1)

The method of claim 1, wherein the plurality of data objects comprise a first data object, wherein the method further comprises sending, by the compute node after performing data access on the first data object, a metadata update request to the memory node, wherein the metadata update request instructs to update access information of the first data object, and wherein metadata of the cache data comprises the access information.

### Claim 4 (depends on 2)

The method of claim 2, wherein the metadata update request comprises:
a write operation request instructing to set a first value of a last access timestamp field in metadata of the first data object to a timestamp at which the compute node initiates the data access, wherein the last access timestamp field indicates a moment at which a data object to which the metadata belongs is accessed last time; and
an atomic counting operation request instructing to increase a second value of a quantity of access times field in the metadata by 1.

### Claim 5 (depends on 1)

The method of claim 1, wherein the cold data comprises a second data object, wherein after sending the cache replacement request to the memory node, the method further comprises sending a historical record entry generation request to the memory node, wherein the historical record entry generation request instructs to convert metadata of the second data object into a historical record entry, and wherein the historical record entry indicates a second cache replacement algorithm used when performing cache replacement on the second data object.

### Claim 6 (depends on 5)

The method of claim 5, wherein the historical record entry generation request comprises:
an atomic swapping operation request instructing to modify a first value of a length field in the metadata to a second value of a type field, wherein the type field indicates performance of cache replacement on the second data object; and
first write operation request instructing to replace a third value of a field that is in the metadata and that is used for representing the access information with a fourth value representing the second cache replacement algorithm.

### Claim 7 (depends on 6)

The method of claim 6, further comprising:
determining, when access to the second data object stored in the memory node is a miss, that the memory node stores the historical record entry of the second data object;
sending, to the memory node, a second read operation request for reading the historical record entry of the second data object from the memory node, wherein a fifth value of a second algorithm identifier field of the historical record entry of the second data object indicates the second cache replacement algorithm; and
reducing a weight of the second cache replacement algorithm.

### Claim 8 (depends on 7)

The method of claim 7, further comprising:
sending a second write operation request to the memory node, wherein the second write operation request instructs to write a reduced weight of the second cache replacement algorithm into the memory node such that the memory node combines the reduced weight of the second cache replacement algorithm into a combined global algorithm weight; and
receiving, at the compute node, the combined global algorithm weight from the memory node.

### Claim 9 (depends on 1)

The method of claim 1, wherein sending the cache replacement request to the memory node comprises sending the cache replacement request to the memory node when a remaining storage space of the memory node is less than a size of a to-be-inserted data object or less than a preset threshold.

### Claim 10

A computer program product comprising instructions that are stored on a non-transitory computer-readable medium and that, when executed by a processor, cause a compute node of a disaggregated memory architecture to:
send a first read operation request to a memory node of the disaggregated memory architecture, wherein the first read operation request requests reading access information of a plurality of data objects from the memory node, and wherein the memory node stores cache data that comprises the plurality of data objects;
receive, from the memory node in response to the first read operation request, the access information;
separately calculate popularities of the plurality of data objects based on the access information using a first cache replacement algorithm in a plurality of cache replacement algorithms; and
send, when the compute node is to perform cache replacement on the cache data, a cache replacement request to the memory node based on the popularities,
wherein the cache replacement request instructs to replace cold data in the plurality of data objects with a data object to be inserted into the memory node.

### Claim 11 (depends on 10)

The computer program product of claim 10, wherein the plurality of data objects comprise a first data object, wherein the instructions, when executed by the processor, further cause the compute node to send, after performing data access on the first data object, a metadata update request to the memory node, wherein the metadata update request instructs to update access information of the first data object, and wherein metadata of the cache data comprises the access information.

### Claim 12 (depends on 11)

The computer program product of claim 11, wherein the metadata update request comprises:
a write operation request instructing to set a first value of a last access timestamp field in metadata of the first data object to a timestamp at which the compute node initiates the data access, wherein the last access timestamp field indicates a moment at which a data object to which the metadata belongs is accessed last time; and
an atomic counting operation request instructing to increase a second value of a quantity of access times field in the metadata by 1.

### Claim 13

A compute node comprising:
a memory configured to store instructions; and
one or more processors coupled to the memory and configured to execute the instructions to cause the compute node to:
send a first read operation request to a memory node, wherein the first read operation request requests reading access information of a plurality of data objects from the memory node, and wherein the memory node stores cache data that comprises the plurality of data objects;
separately calculate popularities of the plurality of data objects based on the access information using a first cache replacement algorithm in a plurality of cache replacement algorithms; and
send, when cache replacement is to be performed on the cache data, a cache replacement request to the memory node based on the popularities of, wherein the cache replacement request instructs to replace cold data in the plurality of data objects with a data object to be inserted into the memory node.

### Claim 14 (depends on 13)

The compute node of claim 13, wherein the plurality of data objects comprise a first data object, wherein the one or more processors are further configured to execute the instructions to cause the compute node to send, after performing data access on the first data object, a metadata update request to the memory node, wherein the metadata update request instructs to update access information of the first data object, and wherein metadata of the cache data comprises the access information.

### Claim 15 (depends on 14)

The compute node of claim 14, wherein the metadata update request comprises:
a write operation request instructing to set a first value of a last access timestamp field in metadata of the first data object to a timestamp at which the compute node initiates the data access, wherein the last access timestamp field indicates a moment at which a data object to which the metadata belongs is accessed last time; and
an atomic counting operation request instructing to increase a second value of a quantity of access times field in the metadata by 1.

### Claim 16 (depends on 13)

The compute node of claim 13, wherein the cold data comprises a second data object, wherein after sending the cache replacement request to the memory node, the one or more processors are further configured to execute the instructions to cause the compute node to send a historical record entry generation request to the memory node, wherein the historical record entry generation request instructs to convert metadata of the second data object into a historical record entry, and wherein the historical record entry indicates a second cache replacement algorithm used when performing cache replacement on the second data object.

### Claim 17 (depends on 16)

The compute node of claim 16, wherein the historical record entry generation request comprises:
an atomic swapping operation request instructing to modify a first value of a length field in the metadata to a second value of a type field, wherein the type field indicates performance of cache replacement on the second data object; and
first write operation request instructing to replace a third value of a field that is in the metadata and that is used for representing the access information with a fourth value representing the second cache replacement algorithm.

### Claim 18 (depends on 17)

The compute node of claim 17, wherein the one or more processors are further configured to execute the instructions to cause the compute node to:
determine, when access to the second data object stored in the memory node is a miss, that the memory node stores the historical record entry of the second data object;
send, to the memory node, a second read operation request for reading the historical record entry of the second data object from the memory node, wherein a fifth value of a second algorithm identifier field of the historical record entry of the second data object indicates the second cache replacement algorithm; and
reduce a weight of the second cache replacement algorithm.

### Claim 19 (depends on 18)

The compute node of claim 18, wherein the one or more processors are further configured to execute the instructions to cause the compute node to:
send a second write operation request to the memory node, wherein the second write operation request instructs to write a reduced weight of the second cache replacement algorithm into the memory node such that the memory node combines the reduced weight of the second cache replacement algorithm into a combined global algorithm weight; and
receive the combined global algorithm weight from the memory node.

### Claim 20 (depends on 13)

The compute node of claim 13, wherein the one or more processors are further configured to execute the instructions to cause the compute node to further send the cache replacement request to the memory node by sending the cache replacement request to the memory node when a remaining storage space of the memory node is less than a size of a to-be-inserted data object or less than a preset threshold.
