Kafka Metrics

Introduction

This section describes metrics reported by DMS for Kafka to Cloud Eye as well as their namespaces and dimensions. You can use the Cloud Eye console or APIs to query the Kafka metrics and alarms, or view Kafka instance metrics on the Monitoring Details page of the DMS for Kafka console.

Cloud Eye supports a maximum of four hierarchical dimensions that are numbered from 0 to 3. For example, if the dimension information of a monitoring metric is "kafka_instance_id,kafka_broker", the dimension of "kafka_instance_id" is numbered 0 and that of "kafka_broker" is numbered 1.

Namespace

SYS.DMS

Instance Metrics

**Table 1** Instance metrics
Metric ID	Metric Name	Description	Value Range	Unit	Conversion Rule	Dimension	Monitoring Period (Raw Data)
current_partitions	Partitions	Number of used partitions in the instance	0–100,000	Count	N/A	kafka_instance_id	1 minute
current_topics	Topics	Number of created topics in the instance	0–100,000	Count	N/A	kafka_instance_id	1 minute
group_msgs	Accumulated Messages	Total number of accumulated messages in all consumer groups of the instance	0–1,000,000,000	Count	N/A	kafka_instance_id	1 minute
instance_bytes_in_rate	Message Production	Number of bytes produced in the instance per second Some instances do not support this metric. Check whether your instance supports it on the console.	0–1,000,000	Byte/s	1024(IEC)	kafka_instance_id	1 minute
instance_bytes_out_rate	Message Consumption	Number of bytes consumed from the instance per second Some instances do not support this metric. Check whether your instance supports it on the console.	0–1,000,000	Byte/s	1024(IEC)	kafka_instance_id	1 minute
current_partitions_usage	Current Partitions Usage	Current Partitions Usage Some instances do not support this metric. Check whether your instance supports it on the console.	0–100	%	N/A	kafka_instance_id	1 minute

Broker Metrics

Enabling Smart Connect for a Kafka instance creates two or more brokers. On the By Broker tab page, select "connector" for Node Type for Smart Connect broker metrics, or select "broker" for Node Type for Kafka instance broker metrics.

Metrics of Smart Connect brokers: disk capacity usage, memory usage, JVM heap memory usage, node alive status, and connections.

**Table 2** Broker metrics
Metric ID	Metric Name	Description	Value Range	Unit	Conversion Rule	Dimension	Monitoring Period (Raw Data)
broker_data_size	Message Size	Total size of messages in the broker	0–5,000,000,000,000	Byte	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_messages_in_rate	Message Creation Rate	Number of messages created per second	0–500,000	Count/s	N/A	kafka_instance_id,kafka_broker	1 minute
broker_bytes_out_rate	Message Retrieval	Number of bytes retrieved per second	0–500,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_bytes_in_rate	Message Creation	Number of bytes created per second	0–500,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_fetch_mean	Average Message Retrieval Processing Duration	Average time that the broker spends processing message retrieval requests	0–10,000	ms	N/A	kafka_instance_id,kafka_broker	1 minute
broker_produce_mean	Average Message Creation Processing Duration	Average time that the broker spends processing message creation requests	0–10,000	ms	N/A	kafka_instance_id,kafka_broker	1 minute
broker_cpu_core_load	Average Load per CPU Core	Average load of each CPU core of the Kafka VM	0–20	N/A	N/A	kafka_instance_id,kafka_broker	1 minute
broker_disk_usage	Disk Capacity Usage	Disk usage of the Kafka VM	0–100	%	N/A	kafka_instance_id,kafka_broker	1 minute
broker_memory_usage	Memory usage of the broker VM	Memory usage of the Kafka VM	0–100	%	N/A	kafka_instance_id,kafka_broker	1 minute
broker_heap_usage	JVM Heap Memory Usage of JVM	Heap memory usage of the Kafka JVM	0–100	%	N/A	kafka_instance_id,kafka_broker	1 minute
broker_alive	Broker Alive	Whether the Kafka broker is alive This metric is supported by instances purchased in April 2020 or later.	1: alive 0: not alive	N/A	N/A	kafka_instance_id,kafka_broker	1 minute
broker_connections	Connections	Total number of TCP connections on the Kafka broker This metric is supported by instances purchased in April 2020 or later.	0–65,535	Count	N/A	kafka_instance_id,kafka_broker	1 minute
broker_cpu_usage	CPU Usage	CPU usage of the Kafka VM This metric is supported by instances purchased in April 2020 or later.	0–100	%	N/A	kafka_instance_id,kafka_broker	1 minute
broker_disk_read_await	Average Disk Read Time	Average time for each disk I/O read in the monitoring period This metric is supported for instances purchased in June 2020 or later.	> 0	ms	N/A	kafka_instance_id,kafka_broker	1 minute
broker_disk_write_await	Average Disk Write Time	Average time for each disk I/O write in the monitoring period This metric is supported for instances purchased in June 2020 or later.	> 0	ms	N/A	kafka_instance_id,kafka_broker	1 minute
broker_total_bytes_in_rate	Inbound Traffic	Inbound traffic per second This metric is supported for instances purchased in June 2020 or later.	0–1,000,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_total_bytes_out_rate	Outbound Traffic	Outbound traffic per second This metric is supported for instances purchased in June 2020 or later.	0–1,000,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_disk_read_rate	Disk Read Speed	Read traffic on the disk This metric is supported for instances purchased on or after May 16, 2022.	≥ 0	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
broker_disk_write_rate	Disk Write Speed	Write traffic on the disk This metric is supported for instances purchased on or after May 16, 2022.	≥ 0	Byte/s	1024(IEC)	kafka_instance_id,kafka_broker	1 minute
network_bandwidth_usage	Network Bandwidth Usage	Network bandwidth usage This metric is supported only: For instances purchased since July 9 2023. For instances purchased before July 9 2023, this metric is supported for brokers if they are added since July 9 2023.	0–100	%	N/A	kafka_instance_id,kafka_broker	1 minute

Topic Metrics

**Table 3** Topic metrics
Metric ID	Metric Name	Description	Value Range	Unit	Conversion Rule	Dimension	Monitoring Period (Raw Data)
topic_bytes_in_rate	Message Creation	Number of bytes created per second This metric is available only when Monitoring Type is set to Basic monitoring on the By Topic tab page.	0–500,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_topics	1 minute
topic_bytes_out_rate	Message Retrieval	Number of bytes retrieved per second This metric is available only when Monitoring Type is set to Basic monitoring on the By Topic tab page.	0–500,000,000	Byte/s	1024(IEC)	kafka_instance_id,kafka_topics	1 minute
topic_data_size	Message Size	Total size of messages in the queue This metric is available only when Monitoring Type is set to Basic monitoring on the By Topic tab page.	0–5,000,000,000,000	Byte	1024(IEC)	kafka_instance_id,kafka_topics	1 minute
topic_messages	Total Messages	Total number of messages in the queue This metric is available only when Monitoring Type is set to Basic monitoring on the By Topic tab page.	≥ 0	Count	N/A	kafka_instance_id,kafka_topics	1 minute
topic_messages_in_rate	Message Creation Rate	Number of messages created per second This metric is available only when Monitoring Type is set to Basic monitoring on the By Topic tab page.	0–500,000	Count/s	N/A	kafka_instance_id,kafka_topics	1 minute
partition_messages	Partition Messages	Total number of messages in the partition This metric is available only when Monitoring Type is set to Partition monitoring on the By Topic tab page.	≥ 0	Count	N/A	kafka_instance_id,kafka_topics,kafka_partitions	1 minute
produced_messages	Created Messages	Number of messages that have been created This metric is available only when Monitoring Type is set to Partition monitoring on the By Topic tab page.	≥ 0	Count	N/A	kafka_instance_id,kafka_topics,kafka_partitions	1 minute
partition_messages_in_rate	Partition message production rate	Number of messages created per second in a partition. This metric is available only when Monitoring Type is set to Partition monitoring on the By Topic tab page. Some instances do not support this metric. Contact customer service.	≥ 0	Count/s	N/A	kafka_instance_id,kafka_topics,kafka_partitions	1 minute

Consumer Group Metrics

**Table 4** Consumer group metrics
Metric ID	Metric Name	Description	Value Range	Unit	Conversion Rule	Dimension	Monitoring Period (Raw Data)
messages_consumed	Retrieved Messages	Number of messages that have been retrieved in the consumer group This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Partition monitoring on the By Consumer Group tab page.	≥ 0	Count	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics,kafka_groups_partitions	1 minute
messages_remained	Number of accumulations Available Messages	Number of messages that can be retrieved in the consumer group This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Partition monitoring on the By Consumer Group tab page.	≥ 0	Count	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics,kafka_groups_partitions	1 minute
topic_messages_remained	Topic Available Messages	Number of remaining messages that can be retrieved from the specified topic in the consumer group This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Basic monitoring on the By Consumer Group tab page.	0 to 2⁶³–1	Count	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics	1 minute
topic_messages_consumed	Topic Retrieved Messages	Number of messages that have been retrieved from the specified topic in the consumer group This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Basic monitoring on the By Consumer Group tab page.	0 to 2⁶³–1	Count	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics	1 minute
consumer_messages_remained	Consumer Available Messages	Number of remaining messages that can be retrieved in the consumer group This metric is available only when Topic is set to All topics on the By Consumer Group tab page.	0 to 2⁶³–1	Count	N/A	kafka_instance_id,kafka_groups	1 minute
consumer_messages_consumed	Consumer Retrieved Messages	Number of messages that have been retrieved in the consumer group This metric is available only when Topic is set to All topics on the By Consumer Group tab page.	0 to 2⁶³–1	Count	N/A	kafka_instance_id,kafka_groups	1 minute
messages_consumed_per_min	Partition Consumption Rate	Number of messages consumed from the specified queue partition in the consumer group every minute This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Partition monitoring on the By Consumer Group tab page. Some instances do not support this metric. Check whether your instance supports it on the console.	0–30,000,000	Count/min	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics,kafka_groups_partitions	1 minute
topic_messages_consumed_per_min	Queue Consumption Rate	Number of messages consumed from the specified queue in the consumer group every minute This metric is available only when Topic is set to a specific topic name and Monitoring Type is set to Basic monitoring on the By Consumer Group tab page. Some instances do not support this metric. Check whether your instance supports it on the console.	0–30,000,000	Count/min	N/A	kafka_instance_id,kafka_groups,kafka_groups_topics	1 minute
consumer_messages_consumed_per_min	Consumer Group Consumption Rate	Number of messages consumed from the consumer group every minute This metric is available only when Topic is set to All topics on the By Consumer Group tab page. Some instances do not support this metric. Check whether your instance supports it on the console.	0–30,000,000	Count/min	N/A	kafka_instance_id,kafka_groups	1 minute

Smart Connect Metrics

Smart Connect metrics are only available for cluster instances.

**Table 5** Smart Connect metrics
Metric ID	Metric Name	Description	Value Range	Unit	Conversion Rule	Dimension	Monitoring Period (Raw Data)
kafka_wait_synchronize_data	Kafka Data to Sync	Data to synchronize in the Kafka migration task	≥ 0	Count	N/A	kafka_instance_id,connector_task	1 minute
kafka_synchronize_rate	Kafka Data Synced per Minute	Data synchronized per minute in the Kafka migration task	≥ 0	Count	N/A	kafka_instance_id,connector_task	1 minute
task_status	Task Status	Status of the current task	0: abnormal 1: normal	N/A	N/A	kafka_instance_id,connector_task	1 minute
message_delay	Message Delay	Time elapsed between when a message is sent from the source and received by the target	≥ 0	ms	N/A	kafka_instance_id,connector_task	1 minute

Precautions:

A Smart Connect task that bidirectionally copies Kafka data is split into two tasks for monitoring: Smart Connect task name_source_0 and Smart Connect task name_source_1.
If all messages in a topic have aged before the next synchronization, there is no Kafka data to be synchronized. However, since the Kafka data synchronization metric uses the offset value that contains aged data, Kafka Data Synced per Minute will display the number of aged messages.

If a monitored object has multiple dimensions, the dimensional level of specific metrics is required when you use APIs to query the metrics.

For example, to query the disk capacity usage (broker_disk_usage) of a Kafka broker, its dimension is "kafka_instance_id,kafka_broker", indicating that kafka_instance_id is numbered 0 and kafka_broker is numbered 1.

To query a single metric by calling the API, the kafka_broker dimension is used as follows:
```
dim.0=kafka_instance_id,6a0df74a-fxxx-497bfdda6c8e&dim.1=kafka_broker,broker-0
```
6a0df74a-fxxx-497bfdda6c8e and broker-0 are the dimension values of kafka_instance_id and kafka_broker. For details about how to obtain them, see the guide in the Dimension table.

To batch query metrics by calling the API, the kafka_broker dimension is used as follows:

"dimensions": [ 
                 { 
                     "name": "kafka_instance_id", 
                     "value": "6a0df74a-fxxx-497bfdda6c8e"    
                 }, 
                 { 
                     "name": "kafka_broker", 
                     "value": "broker-0" 
                 } 
             ]

6a0df74a-fxxx-497bfdda6c8e and broker-0 are the dimension values of kafka_instance_id and kafka_broker. For details about how to obtain them, see the guide in the Dimension table.

Dimension

Key	Value
kafka_instance_id	Kafka instance ID, for example, 6a0df74a-fxxx-497bfdda6c8e. To obtain the value, call the Listing All Instances API and extract the value from the response parameters.
kafka_broker	Kafka instance broker. The broker ID starts from 0 and increases in ascending order. For example, if an instance has three brokers, they are broker-0, broker-1, and broker-2. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.
kafka_topics	Topic name, for example, topic-01. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.
kafka_partitions	Partition number in a topic, for example, 0.
kafka_groups_partitions	Partition number in a topic consumed by a consumer group, for example, 0. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.
kafka_groups_topics	Topic name consumed by a consumer group, for example, topic-01. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.
kafka_groups	Consumer group name of a Kafka instance, for example, group-01. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.
connector_task	Smart Connect task name, for example, smart-connect-01_source_0. To obtain the value, call the Querying Kafka Instance Monitoring Dimensions API and extract the value from the response parameters.