- Release Notes and Announcements
- Product Introduction
- Purchase Guide
- Getting Started
- Operation Guide
- Instance Management
- Creating Instance
- Naming with Consecutive Numeric Suffixes or Designated Pattern String
- Viewing Instance
- Upgrading Instance
- Downgrading Instance Configuration
- Terminating/Returning Instances
- Change from Pay-as-You-Go to Monthly Subscription
- Upgrading Instance Version
- Adding Routing Policy
- Public Network Bandwidth Management
- Connecting to Prometheus
- AZ Migration
- Setting Maintenance Time
- Setting Message Size
- Topic Management
- Consumer Group
- Monitoring and Alarms
- Smart Ops
- Permission Management
- Tag Management
- Querying Message
- Event Center
- Migration to Cloud
- Data Compression
- Instance Management
- CKafka Connector
- Practical Tutorial
- Practical Tutorial of CKafka Client
- Connector Practical Tutorial
- Connecting Flink to CKafka
- Connecting Schema Registry to CKafka
- Connecting Spark Streaming to CKafka
- Connecting Flume to CKafka
- Connecting Kafka Connect to CKafka
- Connecting Storm to CKafka
- Connecting Logstash to CKafka
- Connecting Filebeat to CKafka
- Multi-AZ Deployment
- Log Access
- Replacing Supportive Route (Old)
- Practice Tutorial for Cluster Bandwidth in High CPU Utilization Scenarios
- Practice Tutorial for Cluster Capacity Planning
- Troubleshooting
- API Documentation
- History
- Introduction
- API Category
- Making API Requests
- DataHub APIs
- ACL APIs
- Topic APIs
- DescribeTopicProduceConnection
- BatchModifyGroupOffsets
- BatchModifyTopicAttributes
- CreateConsumer
- CreateDatahubTopic
- CreatePartition
- CreateTopic
- CreateTopicIpWhiteList
- DeleteTopic
- DeleteTopicIpWhiteList
- DescribeDatahubTopic
- DescribeTopic
- DescribeTopicAttributes
- DescribeTopicDetail
- DescribeTopicSubscribeGroup
- FetchMessageByOffset
- FetchMessageListByOffset
- ModifyDatahubTopic
- ModifyTopicAttributes
- DescribeTopicSyncReplica
- Instance APIs
- Route APIs
- Other APIs
- Data Types
- Error Codes
- SDK Documentation
- General References
- Conducting Production and Consumption Pressure Testing on CKafka
- Configuration Guide for Common Parameters in CKafka
- Connecting to Legacy Self-Built Kafka
- Suggestions for CKafka Version Selection
- CKafka Data Reliability Description
- Connector
- FAQs
- Service Level Agreement
- Contact Us
- Glossary
- Release Notes and Announcements
- Product Introduction
- Purchase Guide
- Getting Started
- Operation Guide
- Instance Management
- Creating Instance
- Naming with Consecutive Numeric Suffixes or Designated Pattern String
- Viewing Instance
- Upgrading Instance
- Downgrading Instance Configuration
- Terminating/Returning Instances
- Change from Pay-as-You-Go to Monthly Subscription
- Upgrading Instance Version
- Adding Routing Policy
- Public Network Bandwidth Management
- Connecting to Prometheus
- AZ Migration
- Setting Maintenance Time
- Setting Message Size
- Topic Management
- Consumer Group
- Monitoring and Alarms
- Smart Ops
- Permission Management
- Tag Management
- Querying Message
- Event Center
- Migration to Cloud
- Data Compression
- Instance Management
- CKafka Connector
- Practical Tutorial
- Practical Tutorial of CKafka Client
- Connector Practical Tutorial
- Connecting Flink to CKafka
- Connecting Schema Registry to CKafka
- Connecting Spark Streaming to CKafka
- Connecting Flume to CKafka
- Connecting Kafka Connect to CKafka
- Connecting Storm to CKafka
- Connecting Logstash to CKafka
- Connecting Filebeat to CKafka
- Multi-AZ Deployment
- Log Access
- Replacing Supportive Route (Old)
- Practice Tutorial for Cluster Bandwidth in High CPU Utilization Scenarios
- Practice Tutorial for Cluster Capacity Planning
- Troubleshooting
- API Documentation
- History
- Introduction
- API Category
- Making API Requests
- DataHub APIs
- ACL APIs
- Topic APIs
- DescribeTopicProduceConnection
- BatchModifyGroupOffsets
- BatchModifyTopicAttributes
- CreateConsumer
- CreateDatahubTopic
- CreatePartition
- CreateTopic
- CreateTopicIpWhiteList
- DeleteTopic
- DeleteTopicIpWhiteList
- DescribeDatahubTopic
- DescribeTopic
- DescribeTopicAttributes
- DescribeTopicDetail
- DescribeTopicSubscribeGroup
- FetchMessageByOffset
- FetchMessageListByOffset
- ModifyDatahubTopic
- ModifyTopicAttributes
- DescribeTopicSyncReplica
- Instance APIs
- Route APIs
- Other APIs
- Data Types
- Error Codes
- SDK Documentation
- General References
- Conducting Production and Consumption Pressure Testing on CKafka
- Configuration Guide for Common Parameters in CKafka
- Connecting to Legacy Self-Built Kafka
- Suggestions for CKafka Version Selection
- CKafka Data Reliability Description
- Connector
- FAQs
- Service Level Agreement
- Contact Us
- Glossary
Practice Tutorial for Cluster Capacity Planning
Last updated: 2024-11-07 15:05:18
When you use the TDMQ for CKafka, the main specifications include bandwidth and storage, as well as availability zone (AZ) distribution and the number of partitions. These metrics determine the cluster's load capability to some extent. However, in actual operations, due to differences in business scenarios, the actual load of the cluster may be affected by various factors, such as the message size, whether the messages are compressed, message sending/receiving ratio, number of Topic replicas, and their key attributes. Therefore, it is not comprehensive enough to simply use the cluster's bandwidth and storage ratio as the sole judgment metrics for cluster scaling.
To better ensure the stable operation of business and to plan and manage cluster capacity reasonably, advanced monitoring currently provides the cluster load metric. This metric can help you obtain the current cluster's load status in a simpler way, thus serving as a reference for assessing whether the current CKafka cluster needs to be scaled out.
Applicable Scenarios
CKafka Pro Edition.
In some special business scenarios, the bandwidth utilization is low but the cluster load is high. Therefore, the cluster bandwidth needs to be scaled out based on the cluster load metric.
Metric Viewing Path
View the overall cluster load at the node level. For more information, see Querying Advanced Monitoring (Pro Edition).
Reference Policies
To ensure the stability of your production business and the processing performance of the CKafka cluster, it is recommended to reasonably plan the cluster capacity according to the deployment method and load status of the cluster. If the cluster load exceeds the following reference value, it is recommended to increase the cluster bandwidth specifications promptly.
Single-AZ deployment
When the cluster is deployed in a single AZ, it is recommended that the maximum cluster load value be kept about 70%.
Multi-AZ deployment
When the cluster is deployed in multiple AZs, a certain level of redundancy needs to be considered so that if an unexpected exception occurs in one AZ, the remaining AZs can handle the business load normally. For example:
Two-AZ deployment: When a single AZ is unavailable, the cluster has half of its nodes remaining. Considering the 70% utilization, it is recommended to keep the normal load of the cluster below 35%.
Three-AZ deployment: When a single AZ is unavailable, the cluster has 2/3 of its nodes remaining. Considering the 70% utilization, it is recommended to keep the normal load of the cluster below 47%.
Yes
No
Was this page helpful?