Metric Name | Parameter | Unit | Description | Calculation Method |
CPU Utilization | cpu_util | % | Proxy CPU utilization. | Collect the CPU utilization of Proxy processes in real time. |
Metric Name | Parameter | Unit | Description | Calculation Method |
Total number of requests | proxy_commands | Times/second | Number of commands executed by Proxy. | Read the Proxy command counter. |
Number of key requests | cmd_key_count | Counts/second | Number of keys accessed by the command. | Read the internal Key access counter of Proxy. |
Number of MGET requests | cmd_mget | Times/second | Number of MGET command executions. | Read the Mget command counter of Proxy. |
Execution error | cmd_err | Times/second | Number of Proxy command execution errors (including nonexistent commands and invalid parameters). | Read the error command counter of Proxy. |
Large value requests | cmd_big_value | Times/second | Number of executions of request commands with the size greater than 32 KB. | Read the large Value request counter of Proxy. |
Maximum commands per connection | connect_commands_max | | Maximum number of commands executed per connection. | Read the command counters of all connections within Proxy and obtain the maximum value. |
Metric Name | Parameter | Unit | Description | Calculation Method |
Number of connections | connections | Counts | Number of TCP connections to the instance | Read the current number of TCP connections directly. |
Connection utilization | connections_util | % | Ratio of the actual TCP connections to the maximum number of connections. | Current connections / maximum connections × 100% |
Maximum Connection Utilization of the Node | connections_max_util | % | Maximum value of the connection ratio of each node among all Proxy nodes. | MAX(connections of each Proxy node / maximum connections of the node) × 100% |
Inbound traffic | in_flow | Mb/s | Traffic flowing into the database per second. | Read the inbound traffic counter. |
Inbound traffic utilization | in_bandwidth_util | % | The ratio of the actually used private network inbound traffic to the maximum bandwidth | Inbound traffic / maximum inbound bandwidth × 100% |
Throttling-triggered inbound traffic | in_flow_limit | | Number of times when the inbound traffic exceeds the maximum bandwidth. | Read the inbound traffic throttling trigger counter. |
Outbound traffic | out_flow | Mb/s | Traffic flowing out of the database per second. | Read the outbound traffic counter. |
Outbound traffic utilization | out_bandwidth_util | % | The ratio of actual outbound traffic usage to maximum bandwidth | Outbound traffic / maximum outbound bandwidth × 100% |
Outbound traffic throttling trigger | out_flow_limit | | Number of times when outbound traffic exceeds the maximum bandwidth. | Read the outbound traffic throttling trigger counter. |
Metric Name | Parameter | Unit | Description | Calculation Method |
Average execution latency | latency_avg | ms | Average execution latency from Proxy to data nodes | Average execution time of all commands within a collection period |
Maximum execution latency | latency_max | ms | Maximum execution latency from Proxy to data nodes | Maximum execution time of all commands within a collection period |
Average read latency | latency_read | ms | Average execution latency of read commands from Proxy to data nodes | Average execution time of read commands within a collection period |
Average write latency | latency_write | ms | Average execution latency of write commands from Proxy to data nodes | Average execution time of write commands within a collection period |
Average latency of other commands | latency_other | ms | Average execution latency of commands other than read and write commands from Proxy to data nodes | Average execution time of other types of commands within a collection period |
P99 latency | latency_p99 | ms | 99% execution latency from Proxy to data nodes | MAX: MAX(latency_p99 of all Proxy nodes) |
Metric Name | Parameter | Unit | Description | Calculation Method |
CPU Utilization | cpu_util | % | Average CPU utilization. | Directly read the CPU utilization from Redis INFO (real-time status collection). |
Metric Name | Parameter | Unit | Description | Calculation Method |
Number of connections | connections | Counts | Number of connections from Proxy to the node. | Directly read the current number of connections (real-time status collection). |
Connection utilization | connections_util | % | Node connection utilization. | Current connections / maximum connections × 100% |
Metric Name | Parameter | Unit | Description | Calculation Method |
Memory usage | mem_used | MB | Memory capacity actually used, including data and cache. | Directly read used_memory from Redis INFO (real-time status collection). |
Memory Utilization | mem_util | % | Ratio of actual memory usage to total memory requested. | used_memory / maxmemory × 100% |
Memory tilt rate | mem_slope_util | % | Calculate the ratio of memory usage of each shard master node to the average memory usage of all shard master nodes. | Node mem_used / AVG (mem_used of all master nodes) × 100%Note: If the monitoring metric is greater than 100%, it means that the node is tilted. |
Total number of keys | keys | Counts | Total number of keys stored in an instance (first-level keys). | Directly read db0:keys from Redis INFO (real-time status collection). |
Number of expired keys | expired | Counts | Number of keys evicted within the time window | Incremental calculation: Current expired_keys - Previous expired_keys |
Number of evicted keys | evicted | Counts | Number of keys evicted within the time window | Incremental calculation: Current evicted_keys - Previous evicted_keys |
The number of keys with expiration time set | expires | Counts | The number of keys (primary key) with expiration time set in the instance | Directly read db0:expires from Redis INFO (real-time status collection). |
Replication delay | repl_delay | Byte | The relative command delay length of the replica node to the primary node | Conditional collection (collected on Master only, attributed to Slave): master_repl_offset - slave_offset |
Metric Name | Parameter | Unit | Description | Calculation Method |
Total number of requests | commands | Times/second | Number of command executions (QPS). | Rate calculation: (current total_commands_processed - previous value) / collection interval |
Read request | cmd_read | Times/second | Number of read command executions. | Rate calculation: (current cumulative read commands - previous value) / collection interval |
Write request | cmd_write | Times/second | Number of write command executions. | Rate calculation: (current cumulative write commands - previous value) / collection interval |
Number of other requests | cmd_other | Times/second | Number of executions of commands other than read and write commands. | Rate calculation: (current cumulative other commands - previous value) / collection interval |
Total request tilt rate | qps_slope_util | % | Calculate the ratio of the number of requests of each shard to the average number of requests of nodes participating in aggregation within the instance. | Node request volume / AVG (request volume of nodes involved in aggregation) × 100%Note: If read-only replicas are enabled, the statistics will include the number of read requests from read-only replica nodes. If the monitoring metric is greater than 100%, it means that the node is tilted. |
Metric Name | Parameter | Unit | Description | Calculation Method |
Slow query | cmd_slow | | Number of command requests with the execution latency greater than the value of slowlog-log-slower-than. | Incremental calculation: Current cumulative slow query count - Previous cumulative slow query count |
Read request hits | cmd_hits | | Number of read requests where the Key exists (keyspace_hits) | Incremental calculation: Current keyspace_hits - Previous keyspace_hits |
Read request miss | cmd_miss | | Number of read requests where the Key does not exist (keyspace_misses) | Incremental calculation: Current keyspace_misses - Previous keyspace_misses |
Read request hit rate | cmd_hits_ratio | % | Cache hit rate | Derived calculation: cmd_hits / (cmd_hits + cmd_miss) × 100%. When the number of accesses is 0, the value is null. |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
CPU Utilization | cpu_util | % | Average CPU utilization of all data nodes (including master and replica nodes) in the instance | AVG: SUM(cpu_util of all data nodes) / total number of data nodes (including master nodes + replica nodes) |
Maximum CPU utilization of nodes | cpu_max_util | % | Maximum CPU utilization of nodes (shards or replicas) in the instance | MAX: MAX(cpu_util of all data nodes) |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
Memory usage | mem_used | MB | Memory capacity actually used, including data and cache | SUM: SUM(mem_used of all master nodes) |
Memory Utilization | mem_util | % | Ratio of actual memory usage to total memory requested. | SUM(mem_used of all master nodes) / instance maxmemory × 100% |
Maximum memory utilization of nodes | mem_max_util | % | Maximum memory utilization of nodes (shards or replicas) in the instance | MAX: MAX(mem_util of all data nodes) |
Total number of keys | keys | Counts | Total number of keys stored in an instance (first-level keys). | SUM: SUM(keys of all master nodes) |
Number of expired keys | expired | Counts | Number of keys evicted within the time window | SUM: SUM(expired of all master nodes) |
Number of evicted keys | evicted | Counts | Number of keys evicted within the time window | SUM: SUM(evicted of all master nodes) |
The number of keys with expiration time set | expires | Counts | The number of keys (primary key) with expiration time set in the instance | SUM: SUM(expires of all master nodes) |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
Number of connections | connections | Counts | Number of TCP connections to the instance | SUM: SUM(connections of all Proxy nodes) |
Connection utilization | connections_util | % | Ratio of the actual number of TCP connections to the maximum number of connections | SUM(connections of all Proxy nodes) / Maximum connections of the instance × 100% |
Maximum Connection Utilization of the Node | connections_max_util | % | Maximum value of the connection ratio of each node among all Proxy nodes. | MAX: MAX(connections of each Proxy node / maximum connections of the node) × 100% |
Inbound traffic | in_flow | Mb/s | Traffic flowing into the database per second. | SUM: SUM(in_flow of all Proxy nodes) |
Inbound traffic utilization | in_bandwidth_util | % | The ratio of the actually used private network inbound traffic to the maximum bandwidth | SUM(in_flow of all Proxy nodes) / Maximum inbound bandwidth of the instance × 100% |
Maximum inbound bandwidth utilization of the node | in_bandwidth_max_util | % | Maximum value of the inbound traffic utilization of each node among all Proxy nodes. | MAX: MAX(in_flow of each Proxy node / maximum inbound bandwidth of the node) × 100% |
Throttling-triggered inbound traffic | in_flow_limit | | Number of times when the inbound traffic exceeds the maximum bandwidth. | SUM: SUM(in_flow_limit of all Proxy nodes) |
Outbound traffic | out_flow | Mb/s | Traffic flowing out of the database per second. | SUM: SUM(out_flow of all Proxy nodes) |
Outbound traffic utilization | out_bandwidth_util | % | The ratio of actual outbound traffic usage to maximum bandwidth | SUM(out_flow of all Proxy nodes) / Maximum outbound bandwidth of the instance × 100% |
Maximum outbound bandwidth utilization of the node | out_bandwidth_max_util | % | Maximum value of the outbound traffic utilization of each node among all Proxy nodes. | MAX: MAX(out_flow of each Proxy node / maximum outbound bandwidth of the node) × 100% |
Outbound traffic throttling trigger | out_flow_limit | | Number of times when outbound traffic exceeds the maximum bandwidth. | SUM: SUM(out_flow_limit of all Proxy nodes) |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
Average execution latency | latency_avg | ms | Average execution latency from Proxy to data nodes | AVG: SUM(latency_avg of all Proxy nodes) / Number of Proxy nodes |
Maximum execution latency | latency_max | ms | Maximum execution latency from Proxy to data nodes | MAX: MAX(latency_max of all Proxy nodes) |
Average read latency | latency_read | ms | Average execution latency of read commands from Proxy to data nodes | AVG: SUM(latency_read of all Proxy nodes) / Number of Proxy nodes |
Average write latency | latency_write | ms | Average execution latency of write commands from Proxy to data nodes | AVG: SUM(latency_write of all Proxy nodes) / Number of Proxy nodes |
Average latency of other commands | latency_other | ms | Average execution latency of commands other than read and write commands from Proxy to data nodes | AVG: SUM(latency_other of all Proxy nodes) / Number of Proxy nodes |
P99 latency | latency_p99 | ms | 99% execution latency from Proxy to data nodes | MAX: MAX(latency_p99 of all Proxy nodes) |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
Total number of requests | commands | Times/second | Number of command executions (QPS). | SUM: SUM(commands of all master nodes + cmd_read of all replica nodes) |
Read request | cmd_read | Times/second | Number of read command executions. | SUM: SUM(cmd_read of all master nodes + cmd_read of all replica nodes) |
Write request | cmd_write | Times/second | Number of write command executions. | SUM: SUM(cmd_write of all master nodes) |
Number of other requests | cmd_other | Times/second | Number of executions of commands other than read and write commands. | SUM: SUM(cmd_other of all master nodes) |
Large value requests | cmd_big_value | Times/second | Number of executions of request commands with the size greater than 32 KB. | SUM: SUM(cmd_big_value of all Proxy nodes) |
Number of key requests | cmd_key_count | Counts/second | Number of keys accessed by the command. | SUM: SUM(cmd_key_count of all Proxy nodes) |
Number of MGET requests | cmd_mget | Counts/second | Number of MGET command executions. | SUM: SUM(cmd_mget of all Proxy nodes) |
Slow query | cmd_slow | | Number of commands with the execution latency greater than the value of slowlog-log-slower-than | SUM: SUM(cmd_slow of all master nodes + cmd_slow of all replica nodes) |
Read request hits | cmd_hits | | Number of read requests where the Key exists (keyspace_hits) | SUM: SUM(cmd_hits of all master nodes + cmd_hits of all replica nodes) |
Read request miss | cmd_miss | | Number of read requests where the Key does not exist (keyspace_misses) | SUM: SUM(cmd_miss of all master nodes + cmd_miss of all replica nodes) |
Execution error | cmd_err | | Number of command execution errors. | SUM: SUM(cmd_err of all Proxy nodes) |
Read request hit rate | cmd_hits_ratio | % | Total number of read request Key hits across all data nodes / (total hits + total misses). When the number of accesses is 0, this value is null. | Weighted average: SUM(cmd_hits of all nodes) / (SUM(cmd_hits of all nodes) + SUM(cmd_miss of all nodes)) × 100% |
Metric Name | Parameter | Unit | Description | Aggregation Algorithm and Calculation Method |
Slow query | cmd_slow | | Number of commands with the execution latency greater than the value of slowlog-log-slower-than | SUM: SUM(cmd_slow of all master nodes + cmd_slow of all replica nodes) |
Read request hits | cmd_hits | | Number of read requests where the Key exists (keyspace_hits) | SUM: SUM(cmd_hits of all master nodes + cmd_hits of all replica nodes) |
Read request miss | cmd_miss | | Number of read requests where the Key does not exist (keyspace_misses) | SUM: SUM(cmd_miss of all master nodes + cmd_miss of all replica nodes) |
Execution error | cmd_err | | Number of command execution errors. | SUM: SUM(cmd_err of all Proxy nodes) |
Read request hit rate | cmd_hits_ratio | % | Total number of read request Key hits across all data nodes / (total hits + total misses). When the access quantity is 0, this value is null. | Weighted average: SUM(cmd_hits of all nodes) / (SUM(cmd_hits of all nodes) + SUM(cmd_miss of all nodes)) × 100% |
피드백