Item | Feature | CKafka | Self-built Kafka |
Basic feature | Configuring topic parameters | Multiple parameters can be adjusted in the console, including number of partitions, number of replicas, message cleanup policy, minimum number of synchronous replicas, segment rolling duration, and message retention period. | CLI-based configuration makes it difficult for service personnel to customize operations. |
| Changing the number of partitions for a topic | The number of partitions for a topic can be changed in the console. | CLI-based configuration makes it difficult for service personnel to customize operations. |
| Sending messages | Messages can be sent in the console. | Not supported. |
| Querying messages | Messages can be queried by offset or time in the console. | CLI-based message consumption is supported, but messages cannot be queried by offset or time. |
| Resetting a consumption offset | The consumption offset can be reset in the console. | The offset is manually configured in the backend, which is cumbersome and error-prone. |
| Management capabilities | Visualized management of clusters, topics, and consumer groups is supported, and more fine-grained management capabilities are provided by using tags. Statistical analysis of production and consumption traffic of topics is supported, and traffic details can be viewed. | An open-source management system is required, resulting in poor usability. |
Intelligent Ops | Dynamic partitioning | Manual partition balancing is supported.In case of partition skew, partitions are reallocated to nodes. Automatic partition balancing is supported. Partition distribution of topics is automatically checked at the user-defined time, and partition balancing is automated during off-peak hours. | Not supported. |
| Intelligent inspection | The one-click diagnosis feature is provided, which can proactively identify cluster issues and risks, provide solutions based on the experience of Tencent Cloud experts, and automatically summarize health check results to generate reports, delivering a closed-loop Ops experience. | Not supported. |
| Intelligent disk configuration | Automatic disk watermark adjustment is supported. When the disk watermark reaches the threshold, the message retention period is automatically adjusted, or the disk capacity is automatically expanded to prevent impact on businesses. | Not supported. |
Isolation | Multi-instance | Multiple instances can be created based on a single physical cluster. | Not supported. |
| Bandwidth quota control | Bandwidth-based usage limitation of each instance is supported. | Not supported. |
| Topic traffic throttling | Topic traffic throttling is supported to maximize service reliability. | Not supported. |
| Disk quota control | Disk-based usage limitation of each instance is supported. | Not supported. |
| Multi-access point configuration | Routing policies for instances can be configured, and access points can be set in the specified Virtual Private Cloud (VPC) network. | Not supported. The broker address needs to be specified for message sending, and ZooKeeper is required for coordinating and scheduling message receiving, resulting in low flexibility. |
| Isolation of the control plane from the data plane | Architecturally, the control plane is isolated from the data plane. Control operations are implemented by using TencentCloud API SDK to call control components. Data plane operations are performed by using the compatible Apache Kafka SDK. | Not supported. |
| Root account isolation | Logical isolation of data between root accounts is implemented based on the root account and sub-account system of Tencent Cloud. | Not supported. |
Monitoring and alarms | Out-of-the-box | Mature and standardized deployment and monitoring solutions are out-of-the-box. | Not supported. Users need to build and maintain an open-source monitoring system. Furthermore, the system cannot be easily customized if it does not meet the requirements. |
| Monitoring and alarms | Many monitoring metrics are provided, including peak production and consumption bandwidth, average production and consumption bandwidth, number of produced and consumed messages, disk utilization, number of instance connections, and number of unconsumed messages. Multi-dimensional monitoring and alarms for instances, topics, and consumer groups are supported. | Not supported. Third-party open-source components are to be connected. |
| Metric ranking | You can view the top 10 topics by production and consumption traffic, top 10 topics by disk usage, and top 10 topics by consumption speed of consumer groups. | Not supported. |
High availability | Cross-AZ high-availability deployment | Cross-availability zone (AZ) high-availability deployment is supported, and mature fault recovery solutions are provided. Single-AZ deployment can be upgraded to cross-AZ deployment. | Supported but cumbersome. You need to design the deployment solutions and parameters. |
| Cross-region disaster recovery | Data synchronization between instances and replication/migration between instances in different regions are supported. All instance data and metadata will be synchronized. | Supported but cumbersome. The open source solutions are defective. |
Security and compliance | ACL access control of topics | Access control based on username and IP range is supported. Access control of instances and topics can be configured. Also, both blocklist and allowlist modes are supported. | Parameters are configured by using the CLI, which is cumbersome and error-prone. |
| SASL password authentication | SASL authentication can be configured in the console. | Parameters are configured by using the CLI, which is cumbersome and error-prone. |
| SSL authentication | SSL authentication can be configured in the console. | Parameters are configured by using the CLI, which is cumbersome and error-prone. |
| Control operation permission control | Console operations are based on Tencent Cloud Access Management (CAM) control permissions, ensuring security and compliance. | Fine-grained control of the control operation permissions is challenging. |
| Operation traceability | Control operations are interconnected with CloudAudit for traceability. | Not supported. |
Usage habits | Naming specifications | Topics: A topic name can contain a maximum of 128 characters, including only letters, digits, underscores, hyphens (-), and periods (.), and cannot start with double underscores. Consumer groups: A topic name can contain a maximum of 128 characters, including only letters, digits, underscores, hyphens (-), and periods (.). Users: A user name can contain a maximum of 64 characters, including only letters, digits, underscores, hyphens (-), and periods (.). Note: It is recommended to name resources according to the above specifications. Excessively long resource names or those containing special characters may be subject to restrictions during transmission to other systems. | Topics: A topic name can contain a maximum of 249 characters. Consumer groups and users: no limitations. |
Others | Online upgrade capability | CKafka versions can be seamlessly upgraded in case of bugs or security vulnerabilities in community editions. | Not supported. A manual upgrade is required, which may affect services. |
Feedback