Real-time Monitoring and Alerting Mechanism of Doris Database

# 1. Overview of Doris Database Monitoring

Doris database monitoring is crucial for ensuring the stability and efficiency of the database. Through monitoring, we can stay informed about the database's operational status in real time, promptly identify and resolve issues, and safeguard the smooth running of business operations. Doris database monitoring is primarily divided into two categories: system-level monitoring and query-level monitoring.

System-level monitoring focuses on the overall operational status of the database, including metrics such as CPU utilization, memory usage, and disk I/O. These metrics can reflect the consumption of resources and overall performance of the database. Query-level monitoring, on the other hand, concentrates on the execution status of individual queries, encompassing metrics like query latency, query concurrency, and query error rate. Through query-level monitoring, we can assess the efficiency of query execution and detect any anomalies.

# 2. Doris Database Monitoring Metrics

## 2.1 System-Level Monitoring Metrics

System-level monitoring metrics reflect the overall operational status of the Doris database, including CPU utilization, memory usage, and disk I/O.

### 2.1.1 CPU Utilization

CPU utilization indicates the usage of CPU resources. High CPU utilization may lead to query delays and system instability.

SELECT * FROM system.cpu_usage;

**Parameter Description:**

* `user_cpu_usage`: User-mode CPU usage
* `system_cpu_usage`: Kernel-mode CPU usage
* `total_cpu_usage`: Overall CPU usage

**Logical Analysis:**

* `user_cpu_usage` reflects the CPU time consumed by user processes.
* `system_cpu_usage` reflects the CPU time consumed by kernel processes.
* `total_cpu_usage` indicates the overall CPU usage.

### 2.1.2 Memory Usage

Memory usage reflects the consumption of memory resources. High memory usage may result in decreased system performance and OOM (Out Of Memory) errors.

SELECT * FROM system.mem_usage;

**Parameter Description:**

* `total_mem_usage`: Total memory usage
* `used_mem_usage`: Used memory usage
* `free_mem_usage`: Free memory usage

**Logical Analysis:**

* `total_mem_usage` reflects the total physical memory size of the system.
* `used_mem_usage` reflects the used physical memory size of the system.
* `free_mem_usage` reflects the remaining physical memory size of the system.

### 2.1.3 Disk I/O

Disk I/O reflects the performance of disk read/write operations. High disk I/O may lead to query delays and system instability.

SELECT * FROM system.disk_io;

**Parameter Description:**

* `read_iops`: Number of disk read operations per second
* `write_iops`: Number of disk write operations per second
* `read_throughput`: Data volume read per second
* `write_throughput`: Data volume written per second

**Logical Analysis:**

* `read_iops` reflects the number of read operations per second.
* `write_iops` reflects the number of write operations per second.
* `read_throughput` reflects the data volume read per second.
* `write_throughput` reflects the data volume written per second.

## 2.2 Query-Level Monitoring Metrics

Query-level monitoring metrics reflect the performance of individual queries, including query latency, query concurrency, and query error rate.

### 2.2.1 Query Latency

Query latency indicates the time taken from query submission to completion. High query latency may result in poor user experience and decreased system performance.

SELECT * FROM system.query_latency;

**Parameter Description:**

* `query_id`: Query ID
* `start_time`: Query start time
* `end_time`: Query end time
* `latency`: Query latency

**Logical Analysis:**

* `query_id` identifies the query.
* `start_time` and `end_time` reflect the time range of query execution.
* `latency` indicates the time taken from query submission to completion.

### 2.2.2 Query Concurrency

Query concurrency indicates the number of queries executing simultaneously. High query concurrency may lead to resource contention and decreased performance.

S