DocsICP FilingConsole
Documentation
Home Page ApsaraDB RDS Developer Reference RDS API Reference Appendix Performance parameters

Performance parameters

Updated at:
Copy as MD
Product
My Favorites

The DescribeDBInstancePerformance operation queries performance data for an RDS instance. Each call requires a time range (StartTime and EndTime), an instance ID (DBInstanceId), and one or more performance metric keys (Key). Valid Key values and their returned metrics differ by database engine.

ApsaraDB RDS for MySQL and ApsaraDB RDS for MariaDB

ApsaraDB RDS for MySQL and ApsaraDB RDS for MariaDB use the same performance parameter names.

Resource utilization

Key Returned metrics and descriptions
MySQL_MemCpuUsage (non-serverless instances) CPU utilization and memory usage relative to the operating system. Unit: %.
MySQL_RCU_MemCpuUsage (serverless instances) Serverless instance CPU usage and MySQL instance memory usage as a percentage of total operating system resources. Unit: %.
MySQL_IOPS Input/output operations per second (IOPS). Unit: IOPS.
MySQL_MBPS Read and write throughput. Unit: bytes/s.
MySQL_NetworkTraffic Average inbound traffic and average outbound traffic. Unit: KB/s.
MySQL_DetailedSpaceUsage Total, data, log, temporary, and system space usage. Unit: MB.

Query performance

Key Returned metrics and descriptions
MySQL_QPSTPS Average SQL statements executed per second (executions/s) and average transactions per second (TPS).
MySQL_COMDML Average executions per second for DELETE, INSERT, INSERT_SELECT, REPLACE, REPLACE_SELECT, SELECT, and UPDATE statements. Unit: executions/s.
MySQL_SelectScan Number of full table scans. Unit: scans.
MySQL_TempDiskTableCreates Number of temporary tables automatically created on disk during statement execution. Unit: tables.

Connections and threads

Key Returned metrics and descriptions
MySQL_Sessions Number of active connections and total connections. Unit: connections.
MySQL_ThreadStatus Number of active threads and connected threads. Unit: threads.

InnoDB engine

Key Returned metrics and descriptions
MySQL_InnoDBBufferRatio Read hit ratio, usage, and percentage of dirty blocks for the InnoDB buffer pool. Unit: %.
MySQL_InnoDBDataReadWriten Average data read from and written to the InnoDB storage engine per second. Unit: KB.
MySQL_InnoDBLogRequests Average reads from and writes to the InnoDB buffer pool per second. Unit: reads/s or writes/s.
MySQL_InnoDBLogWrites Average log write requests per second, physical writes to the log file per second, and fsync() writes to the log file per second. Unit: writes/s.
MySQL_RowDML Average rows read from, updated in, deleted from, and inserted into InnoDB tables per second, plus physical writes to the log file per second. Unit: rows/s or writes/s.
MySQL_ROW_LOCK Maximum wait time for a row-level lock (ms), average wait time for a row-level lock (ms), and number of waits for row-level locks.

MyISAM engine

Key Returned metrics and descriptions
MySQL_MyISAMKeyBufferRatio Average usage, read hit ratio, and write hit ratio of the MyISAM key buffer per second. Unit: %.
MySQL_MyISAMKeyReadWrites Average reads from and writes to the MyISAM buffer per second, and average reads from and writes to disk per second. Unit: reads/s or writes/s.

Replication

Key Returned metrics and descriptions
slavestat Replication delay of the read-only instance. Unit: seconds.
MySQL_ReplicationDelay Replication delay of the secondary instance. Unit: seconds.
MySQL_ReplicationThread Status of the I/O replication thread and SQL replication thread. 1 indicates normal. 0 indicates the thread is lost.

ApsaraDB RDS for SQL Server

If a single Key corresponds to multiple metrics, the returned values are separated by &.

Resource utilization

Key Returned metric Description
SQLServer_InstanceCPUUsage sqlserver_inscpu CPU utilization relative to the operating system.
SQLServer_InstanceMemUsage sqlserver_insmem Memory usage of the instance.
SQLServer_IOPS db_io_read Number of read I/O requests per second.
db_io Number of I/O requests per second.
db_io_write Number of write I/O requests per second.
iops_of_ecs IOPS upper limit on the compute side.
iops_of_disk IOPS upper limit on the storage side.
iops_no_burst IOPS upper limit with I/O burst disabled.
iops_burst IOPS upper limit with I/O burst enabled.
SQLServer_MBPS mBps_total Disk I/O throughput per second.
mBps_read Disk read I/O throughput per second.
mBps_write Disk write I/O throughput per second.
mBps_of_ecs Disk I/O throughput upper limit on the compute side.
mBps_of_disk Disk I/O throughput upper limit on the storage side.
mBps_no_burst Disk I/O throughput upper limit with I/O burst disabled.
mBps_burst Disk I/O throughput upper limit with I/O burst enabled.
SQLServer_NetworkTraffic read_kb Average outbound traffic per second.
write_kb Average inbound traffic per second.
SQLServer_InstanceDiskUsage InstanceDiskUsage Disk space usage.
SQLServer_DetailedSpaceUsage ins_size Total space usage.
log_size Log space usage.
data_size Data space usage.
other_size System space usage.
tmp_size Temporary space usage.
disk_mastersize Space used by the SQL Server system database master.
disk_modelsize Space used by the model database, which serves as the template for all new databases on the instance.
disk_msdbsize Space used by the msdb database, which stores SQL Server Agent jobs, backup history, and related data.
disk_errorlog Space occupied by error log files. Clear error logs in the console or via API.
disk_xel Space occupied by Extended Event log files.
disk_trace Space used by SQL Server trace files, such as SQL Profiler trace files.
disk_dump Space occupied by dump files generated when the SQL Server kernel encounters an exception.
SQLServer_RCU rcu RDS Capacity Unit (RCU).
SQLServer_InstanceIOPSUsage InstanceIOPSUsageBurst Actual IOPS usage.
InstanceIOPSUsageNoBurst IOPS usage with I/O burst disabled.
InstanceIOPSBurstRatio IOPS improvement rate after I/O burst is enabled.
SQLServer_InstanceMBPSUsage mBps_usage_no_burst Disk I/O throughput usage with I/O burst disabled.
mBps_usage_burst Actual disk I/O throughput usage.
mBps_burst_ratio Disk I/O throughput improvement rate after I/O burst is enabled.

Query performance

Key Returned metric Description
SQLServer_QPS sqlserver_qps Average SQL statements executed per second.
SQLServer_Transactions sqlserver_transactions Average transactions per second.
sqlserver_write_transactions Average write transactions per second.
SQLServer_FullScans sqlserver_fullscans Average full table scans per second.
SQLServer_Deadlock sqlserver_deadlock Number of deadlocks per second.

SQL compilation

Key Returned metric Description
SQLServer_SQLCompilations sqlserver_sqlcompilations Number of SQL compilations per second.
sqlserver_resqlcompilations Number of statement recompilations per second.
Safe_AutoParams_per_sec Number of safe auto-parameterization attempts per second.
Failed_AutoParams_per_sec Number of failed auto-parameterization attempts per second.
Forced_Parameterizations_per_sec Number of successful forced parameterizations per second.
Unsafe_AutoParams_per_sec Number of unsafe auto-parameterization attempts per second.
AutoParam_Attempts_per_sec Number of auto-parameterization attempts per second.

Sessions and connections

Key Returned metric Description
SQLServer_Sessions Active_Temp_Tables Number of active temporary tables.
Active_Transactions Number of active transactions.
Active_cursors Number of active cursors.
sqlserver_activesession Number of active threads.
sqlserver_connectionreset Total logons initiated from the connection pool per second.
sqlserver_sessions Total number of connections.
SQLServer_Logins sqlserver_logins Number of logons per second.
sqlserver_logout Number of logouts per second.

Locks and waits

Key Returned metric Description
SQLServer_LockWaits Lock_Requests_per_sec Total lock requests received per second.
Lock_Wait_Time_ms Average lock wait time.
sqlserver_lockwaits Number of lock requests waiting per second.
Table_Lock_Escalations_per_sec Number of times locks on a table are escalated to TABLE or HoBT granularity per second.
Average_Latch_Wait_Time Average wait time for latch requests.
Average_Lock_Wait_Time Average wait time for lock requests.
Total_Latch_Wait_Time_ms Total wait time for latches in the last second.
sqlserver_latchwaits Number of latch requests that cannot be granted immediately per second.
Lock_waits Performance statistics of processes waiting for locks.
SQLServer_LockTimeout sqlserver_locktimeout Number of lock requests that time out per second, including NOWAIT lock requests.
SQLServer_Block Wait_for_the_worker Length of the worker thread wait queue.
uncommited_tran Number of uncommitted transactions.
Longest_Transaction_Running_Time Duration of the longest-running transaction.
Processes_blocked Number of blocked processes.

Memory

Key Returned metric Description
SQLServer_MemUsage Lock_Memory_KB Memory occupied by lock structures. Unit: KB.
Connection_Memory_KB Memory occupied by connections. Unit: KB.
Database_Cache_Memory_KB Memory occupied by the database buffer pool. Unit: KB.
Free_Memory_KB Available memory. Unit: KB.
Optimizer_Memory_KB Memory occupied by the query optimizer. Unit: KB.
Total_Server_Memory_KB Total memory currently used by SQL Server. Unit: KB.
SQL_Cache_Memory_KB Memory occupied by the SQL plan cache. Unit: KB.
Stolen_Server_Memory_KB Memory occupied for other purposes. Unit: KB.
SQLServer_AdvancedMemUsage CACHESTORE_SQLCP_KB Memory occupied by the SQL compiled plan cache. Unit: KB.
CACHESTORE_PHDR_KB Memory occupied by the physical driver cache. Unit: KB.
CACHESTORE_OBJCP_KB Memory occupied by the object compiled plan cache. Unit: KB.
MEMORYCLERK_SOSNODE_KB Memory allocated to operating system nodes. Unit: KB.
MEMORYCLERK_SQLCLR_KB Memory occupied by the common language runtime (CLR). Unit: KB.
MEMORYCLERK_SQLSTORENG_KB Memory occupied by the SQL storage engine. Unit: KB.
USERSTORE_SCHEMAMGR_KB Memory occupied by the user schema manager. Unit: KB.
MEMORYCLERK_SOSMEMMANAGER Memory occupied by the operating system memory manager. Unit: KB.
CACHESTORE_SQLCP_VMCommitted_KB SQL plan cache in virtual memory.
CACHESTORE_PHDR_VMCommitted_KB Algebrizer tree temporary memory cache in virtual memory.
CACHESTORE_OBJCP_VMCommitted_KB Object plan cache in virtual memory.
MEMORYCLERK_SOSNODE_VMCommitted_KB SOSNODE (part of SQLOS) in virtual memory.
MEMORYCLERK_SQLCLR_VMCommitted_KB SQLCLR in virtual memory.
MEMORYCLERK_SQLSTORENG_VMCommitted_KB Storage engine component in virtual memory.
USERSTORE_SCHEMAMGR_VMCommitted_KB Metadata of different types of database objects in virtual memory.
MEMORYCLERK_SOSMEMMANAGER_VMCommitted_KB SOSMEMMANAGER (part of SQLOS) in virtual memory.

Buffer and cache

Key Returned metric Description
SQLServer_BufferHit Cache_Hit_Ratio Cache hit ratio.
sqlserver_bufferpool Percentage of pages found in the buffer cache that do not require a disk read.
SQLServer_Cache Cache_Object_Counts Total number of objects in the cache.
Cache_Objects_in_use Number of cache objects currently in use.
Cache_Pages Number of pages occupied by the cache.
SQLServer_PageLife sqlserver_pagelife Average time a cached data page stays in memory.
SQLServer_CheckPoint sqlserver_lazy_writes Number of writes to disk per second.
sqlserver_checkpoint Number of pages written by checkpoints per second.

Pages and I/O

Key Returned metric Description
SQLServer_PagePerf Page_lookups_per_sec Number of page lookups in the buffer pool per second.
sqlserver_pagereads Number of data pages read from disk per second (physical I/O).
sqlserver_pagewrites Number of data pages written to disk per second (physical I/O).
Pages_compressed_per_sec Number of data pages compressed per second.
Page_IO_latch_waits Number of waits to obtain a page I/O latch.
sqlserver_pagesplits Number of page splits per second.
Readahead_pages_per_sec Number of pages on which read-ahead operations are performed per second.
Background_writer_pages_per_sec Number of pages written to disk by the background writer per second.
Page_latch_waits Total number of waits to obtain a page latch.
SQLServer_IndexUsage Index_Searches_per_sec Number of index searches per second.
Forwarded_Records_per_sec Number of times forwarded records are accessed per second.

Transaction log

Key Returned metric Description
SQLServer_LogGrowth Log_Shrinks Number of times the transaction log is shrunk.
Log_Truncations Number of times the transaction log is truncated.
Log_Growths Number of times the transaction log is automatically grown.
SQLServer_LogPerf Log_Bytes_Flushed_per_sec Log bytes flushed to disk per second (physical writes to the transaction log).
Log_Bytes_Received_per_sec Log bytes received per second (transaction log generation rate).
Log_Cache_Reads_per_sec Reads from the log cache per second. These reads do not require disk access.
Log_buffer_waits Number of waits for the log buffer. A high value indicates contention for the log buffer — increase the log buffer size or optimize transactions to resolve it.
Log_Flushes_per_sec Number of times the log is flushed to disk per second. A high value indicates frequent transactions — optimize batch operations if needed.
Log_Pool_Requests_per_sec Number of requests for the log buffer per second. Indicates usage pressure on the log buffer.

Replication and high availability

Key Returned metric Description
SQLServer_AGPerf sqlserver_alwayson_lagsecs Replication lag between the primary and secondary instances in a SQL Server Always On availability group.
sqlserver_alwayson_redoqueuesize_kb Size of the log queue to be redone on the replica. Unit: KB.
sqlserver_alwayson_sendqueuesize_kb Size of the log queue to be sent from the primary replica to the secondary replica. Unit: KB.
Bytes_Received_from_Replica_per_sec Bytes received from the replica per second.
ag_unhealthydb_count Number of unhealthy databases in the availability group.
ag_db_not_in_ag_count Number of databases not in the availability group.
ag_redo_finish_time Timestamp when the replica finishes redoing logs.
Bytes_Sent_to_Replica_per_sec Bytes sent to the replica per second.
Flow_Control_Time_ms_per_sec Duration of flow control per second. Unit: ms.
Flow_Control_per_sec Number of times flow control is triggered per second.
SQLServer_MirrorPerf sqlserver_log_send_queue Log bytes to be sent from the primary replica to the secondary replica.
sqlserver_redo_queue Log bytes to be redone on the replica.
Redo_Bytes_Remaining Log bytes that remain to be redone on the replica.
Redone_Bytes_per_sec Log bytes redone on the replica per second.

Row versioning

Key Returned metric Description
SQLServer_OptimizeConcurrent Version_Cleanup_rate_KB_per_s Size of row versioning store cleaned up per second. Unit: KB.
Version_Generation_rate_KB_per_s Size of row versioning store generated per second. Unit: KB.

Backup

Key Returned metric Description
SQLServer_BackupPerf Backup_per_Restore_Throughput_per_sec Bytes processed per second during a backup or restore operation.

Alarms

Key Returned metric Description
SQLServer_Alarm alarm_abnormal_db Number of databases for which database mirroring is not correctly configured or fails to be synchronized.

ApsaraDB RDS for PostgreSQL

ApsaraDB RDS for PostgreSQL provides an enhanced monitoring feature with finer-grained metrics available in the console. The table below lists the Key values for DescribeDBInstancePerformance and their corresponding enhanced monitoring metrics.

Resource utilization

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
CpuUsage CPU utilization. Unit: %. os.cpu_usage.total.avg
MemoryUsage Memory usage. Unit: %. os.mem_usage.total.avg
PgSQL_IOPS Input/output operations per second (IOPS). Unit: operations/s. (os.iops.data.avg + os.iops.wal.avg) / os.iops.total.avg * 100
PgSQL_SpaceUsage Space usage of the instance. Unit: bytes. os.fs_size.wal_dir.avg + os.fs_size.base_dir.avg (Unit: MB)
PolarDBCPU CPU utilization. os.cpu_usage.sys.avg, os.cpu_usage.user.avg, os.cpu_usage.total.avg
PolarDBMemory Memory usage. os.mem_usage.total.avg
PolarDBLocalIOSTAT `mean_local_iops_read`: Read operations on the data disk per second. `mean_local_iops_write`: Write operations on the data disk per second. os.iops.read.avg, os.iops.write.avg
PolarDBLocalDiskUsage `mean_local_pg_wal_dir_size`: Write-Ahead Logging (WAL) space usage. Unit: MB. `mean_local_pg_log_dir_size`: Log space usage. Unit: MB. `mean_local_base_dir_size`: Data space usage. Unit: MB. os.fs_size.wal_dir.avg, os.fs_size.log_dir.avg, os.fs_size.base_dir.avg

Connections

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
PgSQL_Session Total number of connections. Unit: connections. db.connections.total.avg
PolarDBConnections `mean_active_session`: Number of active connections. `mean_idle_connection`: Number of idle connections. `mean_total_session`: Total number of connections. `mean_waiting_connection`: Number of connections in a waiting state. db.connections.active.avg, db.connections.idle.avg, db.connections.total.avg, db.connections.waiting.avg

Query performance and DML

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
PolarDBRowDML `mean_tup_returned_delta`: Rows scanned per second during full table scans. `mean_tup_fetched_delta`: Rows fetched per second using index scans. `mean_tup_inserted_delta`: Rows inserted per second. `mean_tup_updated_delta`: Rows updated per second. `mean_tup_deleted_delta`: Rows deleted per second. db.sql.tup_returned.avg, db.sql.tup_fetched.avg, db.sql.tup_inserted.avg, db.sql.tup_updated.avg, db.sql.tup_deleted.avg
PolarDBQPSTPS `mean_commits_delta`: Committed transactions per second. `mean_rollbacks_delta`: Rolled-back transactions per second. `mean_deadlocks_delta`: Deadlocks per second. `mean_tps`: Transactions per second. db.transactions.xact_commit.avg, db.transactions.xact_rollback.avg, Deadlock-Free Operation Count, db.transactions.xact_commit.avg + db.transactions.xact_rollback.avg

Long-running queries and transactions

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
PolarDBLongSQL `mean_one_second_executing_sqls`: SQL statements running for more than 1 second. `mean_three_seconds_executing_sqls`: SQL statements running for more than 3 seconds. `mean_five_seconds_executing_sqls`: SQL statements running for more than 5 seconds. db.slow_sql.one_second.avg, db.slow_sql.three_seconds.avg, db.slow_sql.five_seconds.avg
PolarDBLongTransaction `mean_one_second_transactions`: Transactions running for more than 1 second. `mean_three_seconds_transactions`: Transactions running for more than 3 seconds. db.long_transactions.active_one_second.avg, db.long_transactions.active_three_seconds.avg
PolarDBLongIdleTransaction `mean_one_second_idle_transactions`: Transactions idle for more than 1 second. `mean_three_seconds_idle_transactions`: Transactions idle for more than 3 seconds. `mean_five_seconds_idle_transactions`: Transactions idle for more than 5 seconds. db.long_transactions.idle_one_second.avg, db.long_transactions.idle_three_seconds, db.long_transactions.idle_five_seconds.avg
PolarDBLongTwoPCTransaction `mean_one_second_two_pc_transactions`: Two-phase commit (2PC) transactions running for more than 1 second. `mean_three_seconds_two_pc_transactions`: 2PC transactions running for more than 3 seconds. `mean_five_seconds_two_pc_transactions`: 2PC transactions running for more than 5 seconds. db.long_transactions.two_pc_one_second.avg, db.long_transactions.two_pc_three_seconds.avg, db.long_transactions.two_pc_five_seconds.avg

Replication

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
PolarDBReplication `mean_replay_latency_in_mb`: Synchronization latency of the secondary database. Unit: bytes. `mean_send_latency_in_mb`: Send latency of the primary database. Unit: bytes. `mean_logical_rep_latency_in_mb`: Logical replication delay. Unit: bytes. /

Maintenance

Key Returned metrics and descriptions Corresponding enhanced monitoring metric
PolarDBSwellTime Longest bloat duration for garbage collection — garbage older than this duration can be collected. Unit: seconds. db.swell.swell_time.avg

References

Previous：Instance state tableNext：Forbidden keywords