Release Notes and Announcements
- Release Notes
- Announcements
Product Introduction
Billing
- Billing Overview
- Billing Method
- Billable Items
- Free Tier
- Billing Examples
- Viewing and Downloading Bill
- Payment Overdue
- FAQs
Getting Started
- Console
- Getting Started with COSBrowser
User Guide
- Creating Request
- Bucket
- Object
- Data Management
- Batch Operation
- Global Acceleration
- Monitoring and Alarms
- Operations Center
- Data Processing
- Content Moderation
- Smart Toolbox
- Data Processing Workflow
- Application Integration
User Tools
- Tool Overview
- Installation and Configuration of Environment
- COSBrowser
- COSCLI (Beta)
- COSCMD
- COS Migration
- FTP Server
- Hadoop
- COSDistCp
- HDFS TO COS
- GooseFS-Lite Tool
- Online Tools
- Diagnostic Tool
Use Cases
- Overview
- Access Control and Permission Management
- Performance Optimization
- Accessing COS with AWS S3 SDK
- Data Disaster Recovery and Backup
- Domain Name Management Practice
- Image Processing
- Audio/Video Practices
- Workflow
- Direct Data Upload
- Content Moderation
- Data Security
- Data Verification
- Big Data Practice
- COS Cost Optimization Solutions
- Using COS in the Third-party Applications
Migration Guide
Data Lake Storage
- Cloud Native Datalake Storage
- Metadata Accelerator
- GooseFS
Data Processing
- Data Processing Overview
- Image Processing
- Media Processing
- Content Moderation
- File Processing Service
- File Preview
Troubleshooting
- Obtaining RequestId
- Slow Upload over Public Network
- 403 Error for COS Access
- Resource Access Error
- POST Object Common Exceptions
API Documentation
- Introduction
- Common Request Headers
- Common Response Headers
- Error Codes
- Request Signature
- Action List
- Service APIs
- Bucket APIs
- Object APIs
- Batch Operation APIs
- Data Processing APIs
- Job and Workflow
- Content Moderation APIs
- Cloud Antivirus API
SDK Documentation
- SDK Overview
- Preparations
- Android SDK
- C SDK
- C++ SDK
- .NET(C#) SDK
- Flutter SDK
- Go SDK
- iOS SDK
- Java SDK
- JavaScript SDK
- Node.js SDK
- PHP SDK
- Python SDK
- React Native SDK
- Mini Program SDK
- Harmony SDK
- Endpoint SDK Quality Optimization
- Error Codes
Security and Compliance
- Data Disaster Recovery
- Data Security
- Cloud Access Management
FAQs
- Popular Questions
- General
- Billing
- Domain Name Compliance Issues
- Bucket Configuration
- Domain Names and CDN
- Object Operations
- Logging and Monitoring
- Permission Management
- Data Processing
- Data Security
- Pre-signed URL Issues
- SDKs
- Tools
- APIs
Agreements
Contact Us
Glossary

Specifications and Limits

Unduh

Mode fokus

Ukuran font

Terakhir diperbarui: 2026-07-03 10:44:01

Category
Specifications and Limits
Detailed Description
QPS
Limit
For read and write requests: By default, each bucket in Chinese mainland public cloud regions is allocated 30,000 QPS for read requests and 30,000 QPS for write requests. In other regions, each bucket is allocated 3,000 QPS for read requests and 3,000 QPS for write requests by default.
For listing objects in a bucket/listing historical versions/listing multipart uploads in progress: By default, each bucket in all regions shares 1000 QPS.
For creating/deleting/listing buckets: By default, each APPID in all regions is allocated 50 QPS.
For batch delete object requests: By default, each bucket in all regions is allocated 300 QPS. Individual delete requests split from a batch delete operation consume write request QPS.
For origin-pull requests (pull requests triggered after synchronous or asynchronous origin fetching): By default, each bucket in all regions is allocated 200 QPS.
For configuring/deleting custom origin domain names: By default, each bucket in all regions is allocated 1 QPS.
For data retrieval requests: By default, each bucket in all regions is allocated 100 QPS.
For creating a one-time inventory job request: By default, each bucket in all regions is allocated 1 QPS.
For hotspot frequency control on single-file upload/delete/List operations: Each is allocated 50 QPS.
For hotspot frequency control on single-file downloads: 1000 QPS. If you require a higher QPS, see Request Rate and Performance Optimization.
Bandwidth
Limit
For Chinese mainland public cloud regions: By default, the bandwidth for both upload and download is 15 Gbit/s per account per region.
For other regions: By default, the bandwidth for both upload and download is 10 Gbit/s per account per region. If the bandwidth reaches this threshold, traffic control is triggered. If you require higher bandwidth, contact after-sales engineers.
Storage Type
MAZ_STANDARD/MAZ_STANDARD Limit
Billing restrictions:
Storage duration and storage units are unlimited.
For STANDARD pricing details, see Product Pricing.
﻿
MAZ_STANDARD_IA/MAZ_STANDARD_IA Limit
Billing restrictions:
If the storage duration is less than 30 days, it is calculated as 30 days.
If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of STANDARD_IA, see Product Pricing.
﻿
MAZ_INTELLIGENT_TIERING/MAZ_INTELLIGENT_TIERING Limit
Billing restrictions:
Objects smaller than 64 KB are persistently stored in the frequent access tier. Individual storage files are billed based on their actual data size, regardless of their size. For INTELLIGENT_TIERING pricing details, see Product Pricing.
﻿
MAZ_ARCHIVE/MAZ_ARCHIVE Limit
Billing restrictions:
If the storage duration is less than 90 days, it is calculated as 90 days.
If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of ARCHIVE, see Product Pricing.
﻿
DEEP_ARCHIVE Limit
Billing restrictions:
If the storage duration is less than 180 days, it is calculated as 180 days.
If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of DEEP_ARCHIVE, see Product Pricing.
Bucket
Limit
Once a bucket is created, its name and region cannot be modified.
Bucket names must be unique and cannot be renamed within a single user account.
A name cannot start or end with a hyphen ("-"). It can only contain lowercase English letters, digits [a-z, 0-9], hyphens ("-"), and their combinations. The maximum allowed length of a bucket name is affected by the number of characters of the region abbreviation and the APPID. The total number of characters in the fully formed request domain name must not exceed 60.
﻿
 Number of Buckets
Each primary account can have a maximum of 200 (by default).
﻿
 Number of Objects
The number of objects in each bucket is unlimited.
﻿
 Bucket tagging
A bucket can have a maximum of 50 tags, and tag keys cannot be repeated.
Object
Limit
Object key length supports 1 - 850 bytes. For details, see Object Overview.
﻿
Upload
The maximum size for a single object uploaded via the console is 512 GB.
The maximum size for a single object uploaded via API/SDK is 48.82 TB (50,000 GB).                     
Upload Interface Specifications:                        
Simple Upload: The maximum size for a single object is 5 GB. For details, see Simple Upload.
Multipart Upload: The maximum size for a single object is 48.82 TB. The part size ranges from 1 MB to 5 GB, and the last part can be less than 1 MB. The number of parts ranges from 1 to 10,000. For details, see Multipart Upload.
Currently, storage types with multi-AZ features, such as MAZ_STANDARD and MAZ_STANDARD_IA, can be uploaded to buckets that have multi-AZ configuration enabled. If a bucket also has intelligent tiering configuration enabled, the MAZ_INTELLIGENT_TIERING storage type can also be uploaded.
Currently, you can upload INTELLIGENT_TIERING objects only when the INTELLIGENT_TIERING configuration is enabled for a bucket. The transition of objects between different storage tiers is determined by the parameters in the INTELLIGENT_TIERING configuration.
﻿
Copying
A single account can replicate objects within the same region or across regions.
Object replication within the same region is free. Cross-region object replication incurs traffic fees. For details, see the traffic fee information in Billing. 
Replication Interface Specifications:                        
Simple Replication: The maximum size for a single object to be replicated is 5 GB. For details, see Simple Replication.
For objects larger than 5 GB, you must use Multipart Replication. The maximum size for a single object to be replicated is 48.82 TB. For details, see Multipart Replication.
A bucket with Multi-AZ enabled does not support replicating Multi-AZ storage types to Single-AZ storage types.
Currently, replicating STANDARD, STANDARD_IA, or INTELLIGENT_TIERING storage types to the INTELLIGENT_TIERING storage type is not supported.
﻿
Batch Deletion
When batch deletion is initiated via API or SDK, a maximum of 1000 objects can be deleted per operation.
﻿
Object tagging
A maximum of 10 tags can be added to a single object, and tag keys cannot be repeated.
Access Policy
Number of rules.
Each root account (that is, the same APPID) can have a maximum of 1000 bucket ACL rules.
Lifecycle
Number of rules.
Each bucket can have a maximum of 1000 rules.
﻿
Storage type transition
Standard to Infrequent Access: Minimum 1 day.
Standard/Infrequent Access to Archive or Deep Archive: Minimum 1 day.
Note:
Currently, MAZ_STANDARD and MAZ_STANDARD_IA cannot be transitioned to STANDARD_IA, ARCHIVE, or DEEP_ARCHIVE storage types.
Objects smaller than 64 KB are not subject to lifecycle transition operations.
﻿
Deletion upon Expiration
Standard/Infrequent Access/Archive Expiration Deletion: Minimum 1 day.
SDK Types
﻿
15 types: Android, C, C++, .NET, Flutter, Go, iOS, Java, JavaScript, Node.js, PHP, Python, React Native, Mini Program, and Harmony SDK.
API Reserved Fields
﻿
The interface parameters mentioned in the API documentation are all COS reserved fields, including the following parameters: acl, uploads, policy, cors, delete, versions, location, referer, lifecycle, versioning, notification, replication, website, logging, tagging, accelerate, domain, inventory, origin, object-lock, live, encryption, intelligenttiering, symlink, and so on.

Bantuan dan Dukungan

Apakah halaman ini membantu?

Anda juga dapat Menghubungi Penjualan atau Mengirimkan Tiket untuk meminta bantuan.

masukan

tencent cloud

Cloud Object Storage

Specifications and Limits

Bantuan dan Dukungan

Category	Specifications and Limits	Detailed Description
QPS	Limit	For read and write requests: By default, each bucket in Chinese mainland public cloud regions is allocated 30,000 QPS for read requests and 30,000 QPS for write requests. In other regions, each bucket is allocated 3,000 QPS for read requests and 3,000 QPS for write requests by default. For listing objects in a bucket/listing historical versions/listing multipart uploads in progress: By default, each bucket in all regions shares 1000 QPS. For creating/deleting/listing buckets: By default, each APPID in all regions is allocated 50 QPS. For batch delete object requests: By default, each bucket in all regions is allocated 300 QPS. Individual delete requests split from a batch delete operation consume write request QPS. For origin-pull requests (pull requests triggered after synchronous or asynchronous origin fetching): By default, each bucket in all regions is allocated 200 QPS. For configuring/deleting custom origin domain names: By default, each bucket in all regions is allocated 1 QPS. For data retrieval requests: By default, each bucket in all regions is allocated 100 QPS. For creating a one-time inventory job request: By default, each bucket in all regions is allocated 1 QPS. For hotspot frequency control on single-file upload/delete/List operations: Each is allocated 50 QPS. For hotspot frequency control on single-file downloads: 1000 QPS. If you require a higher QPS, see Request Rate and Performance Optimization.
Bandwidth	Limit	For Chinese mainland public cloud regions: By default, the bandwidth for both upload and download is 15 Gbit/s per account per region. For other regions: By default, the bandwidth for both upload and download is 10 Gbit/s per account per region. If the bandwidth reaches this threshold, traffic control is triggered. If you require higher bandwidth, contact after-sales engineers.
Storage Type	MAZ_STANDARD/MAZ_STANDARD Limit	Billing restrictions: Storage duration and storage units are unlimited. For STANDARD pricing details, see Product Pricing.
		MAZ_STANDARD_IA/MAZ_STANDARD_IA Limit	Billing restrictions: If the storage duration is less than 30 days, it is calculated as 30 days. If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of STANDARD_IA, see Product Pricing.
		MAZ_INTELLIGENT_TIERING/MAZ_INTELLIGENT_TIERING Limit	Billing restrictions: Objects smaller than 64 KB are persistently stored in the frequent access tier. Individual storage files are billed based on their actual data size, regardless of their size. For INTELLIGENT_TIERING pricing details, see Product Pricing.
		MAZ_ARCHIVE/MAZ_ARCHIVE Limit	Billing restrictions: If the storage duration is less than 90 days, it is calculated as 90 days. If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of ARCHIVE, see Product Pricing.
		DEEP_ARCHIVE Limit	Billing restrictions: If the storage duration is less than 180 days, it is calculated as 180 days. If a storage unit is less than 64 KB, it is calculated as 64 KB. If a storage unit is greater than or equal to 64 KB, it is calculated based on its actual size. For specific pricing of DEEP_ARCHIVE, see Product Pricing.
Bucket	Limit	Once a bucket is created, its name and region cannot be modified. Bucket names must be unique and cannot be renamed within a single user account. A name cannot start or end with a hyphen ("-"). It can only contain lowercase English letters, digits [a-z, 0-9], hyphens ("-"), and their combinations. The maximum allowed length of a bucket name is affected by the number of characters of the region abbreviation and the APPID. The total number of characters in the fully formed request domain name must not exceed 60.
		Number of Buckets	Each primary account can have a maximum of 200 (by default).
		Number of Objects	The number of objects in each bucket is unlimited.
		Bucket tagging	A bucket can have a maximum of 50 tags, and tag keys cannot be repeated.
Object	Limit	Object key length supports 1 - 850 bytes. For details, see Object Overview.
		Upload	The maximum size for a single object uploaded via the console is 512 GB. The maximum size for a single object uploaded via API/SDK is 48.82 TB (50,000 GB). Upload Interface Specifications: Simple Upload: The maximum size for a single object is 5 GB. For details, see Simple Upload. Multipart Upload: The maximum size for a single object is 48.82 TB. The part size ranges from 1 MB to 5 GB, and the last part can be less than 1 MB. The number of parts ranges from 1 to 10,000. For details, see Multipart Upload. Currently, storage types with multi-AZ features, such as MAZ_STANDARD and MAZ_STANDARD_IA, can be uploaded to buckets that have multi-AZ configuration enabled. If a bucket also has intelligent tiering configuration enabled, the MAZ_INTELLIGENT_TIERING storage type can also be uploaded. Currently, you can upload INTELLIGENT_TIERING objects only when the INTELLIGENT_TIERING configuration is enabled for a bucket. The transition of objects between different storage tiers is determined by the parameters in the INTELLIGENT_TIERING configuration.
		Copying	A single account can replicate objects within the same region or across regions. Object replication within the same region is free. Cross-region object replication incurs traffic fees. For details, see the traffic fee information in Billing. Replication Interface Specifications: Simple Replication: The maximum size for a single object to be replicated is 5 GB. For details, see Simple Replication. For objects larger than 5 GB, you must use Multipart Replication. The maximum size for a single object to be replicated is 48.82 TB. For details, see Multipart Replication. A bucket with Multi-AZ enabled does not support replicating Multi-AZ storage types to Single-AZ storage types. Currently, replicating STANDARD, STANDARD_IA, or INTELLIGENT_TIERING storage types to the INTELLIGENT_TIERING storage type is not supported.
		Batch Deletion	When batch deletion is initiated via API or SDK, a maximum of 1000 objects can be deleted per operation.
		Object tagging	A maximum of 10 tags can be added to a single object, and tag keys cannot be repeated.
Access Policy	Number of rules.	Each root account (that is, the same APPID) can have a maximum of 1000 bucket ACL rules.
Lifecycle	Number of rules.	Each bucket can have a maximum of 1000 rules.
		Storage type transition	Standard to Infrequent Access: Minimum 1 day. Standard/Infrequent Access to Archive or Deep Archive: Minimum 1 day. Note: Currently, MAZ_STANDARD and MAZ_STANDARD_IA cannot be transitioned to STANDARD_IA, ARCHIVE, or DEEP_ARCHIVE storage types. Objects smaller than 64 KB are not subject to lifecycle transition operations.
		Deletion upon Expiration	Standard/Infrequent Access/Archive Expiration Deletion: Minimum 1 day.
SDK Types			15 types: Android, C, C++, .NET, Flutter, Go, iOS, Java, JavaScript, Node.js, PHP, Python, React Native, Mini Program, and Harmony SDK.
API Reserved Fields			The interface parameters mentioned in the API documentation are all COS reserved fields, including the following parameters: acl, uploads, policy, cors, delete, versions, location, referer, lifecycle, versioning, notification, replication, website, logging, tagging, accelerate, domain, inventory, origin, object-lock, live, encryption, intelligenttiering, symlink, and so on.