Monitoring and Alarming

Last updated: 2019-10-25 12:55:47

PDF

The system monitoring of TencentDB for SQL Server supports 25 common parameters of SQL Server. You can collect statistics of other parameters by configuring the counters of SSMS.

Currently Supported Parameters

Common Parameters

Metric Description (Unit) Five-minute Value Optimization Suggestion
CPU utilization
(Total Processor Time)
Actual CPU utilization (%) max < 30% Good
< 60% Good
> 60% Attention required
Number of transactions
(Transaction/sec)
Average transactions per second (times/second) max Dependent on business needs
Number of connections
(User Connection)
Average users connections to the database per second (-) max Dependent on business needs
Number of requests
(Batch Requests/sec)
- (times/second) max Dependent on business needs
Number of logins per second
(Logins/sec)
Logins per second (times/second) max Dependent on business needs
Number of logouts per second
(Logouts/sec)
Logouts per second (times/second) max Dependent on business needs
Used storage space
(Storage Space)
Sum of space occupied by instance database files and log files (GB) max <30% Good
<60% Good
>60% Attention required
Incoming network traffic
(Network Receive Throughput)
Sum of all incoming packets of all connections (MB/s) max Dependent on business needs
Outgoing network traffic
(Network Transmit Throughput)
Sum of all outgoing packets of all connections (MB/s) max Dependent on business needs
Disk IOPS
(IPOS)
Disk reads/writes (times/second) max Follow the suggestions for IOPS in the instance specs center
< 30% Good
< 60% Good
> 60% Attention required
Number of disk reads
(Read IOPS)
Disk reads per second (times/second) max Dependent on business needs
Number of disk writes
(Write IOPS)
Disk writes per second (times/second) max Dependent on business needs

Performance Optimization Parameters

Metric Description (Unit) Five-minute Value Optimization Suggestion
Slow queries
(SlowQuery)
Number of queries running for more than one second (-) avg < 1 Good
< 10 Normal
> 10 Attention required
Blocking processes
(Processes blocked)
Current number of blocking processes (-) avg Dependent on business needs
Number of lock requests
(Lock Requests/sec)
Average lock requests per second (times/second) avg Dependent on business needs
Number of user errors
(User Error/sec)
Average errors per second (times/second) avg 0 Good
> 0 Attention required
Number of lock waits
(Lock waits)
Lock requests requiring caller to wait per second (times/second) avg Dependent on business needs
Number of SQL compilations
(SQL Compilation/sec)
Average SQL compilations per second (times/second) avg Dependent on business needs
Number of SQL recompilations
(SQL Re-Compilation/sec)
Average SQL recompilations per second (times/second) avg Dependent on business needs
Incoming network traffic
(Network Receive Throughput)
Sum of all incoming packets of all connections (MB/s) avg Dependent on business needs
Outgoing network traffic
(Network Transmit Throughput)
Sum of all outgoing packets of all connections (MB/s) avg Dependent on business needs
Number of full-table scans by SQL per second
(Full Scans/sec)
Disk reads/writes (times/second) avg Follow the suggestions for IOPS in the instance specs center
< 30% Good
< 60% Good
> 60% Attention required
Buffer cache hit rate
(Buffer cache hit ratio)
Data cache (memory) hit rate (%) avg >= 95% Good
>= 90% Normal
< 90% Attention required
Number of latch waits per second
(Latch Waits/sec)
Latch waits per second (times/sec) avg Dependent on business needs
Average lock wait delay
(Average Wait Time)
Average wait time for each lock request that causes a wait (ms) max Dependent on business needs
Average network IO delay
(Network IO waits)
Average network IO delay time (ms) max Dependent on business needs
Plan cache hit rate
(Plan Cache: Cache Hit Ratio)
Hit rate of the execution plan of each SQL (%) max >= 95% Good
>= 90% Normal
< 90% Attention required

As SQL Server adopts a full-use mechanism for the memory, there is no need to monitor the direct memory space metric. You can check the memory usage by viewing the cache hit rate.

Currently Supported Alarm Metrics

Currently, alarm is available only for the key performance metrics. For other metrics, alarm is not supported in Cloud Monitor. You can configure the alarm feature in Tencent Cloud’s "Console" -> "Cloud Monitor" -> "My Alarm" -> "Alarm Policy".

Currently, alarming is supported for the following monitoring metrics. The recommended alarm thresholds are as detailed above (in Table 1).

  • CPU utilization
  • Connection count
  • TrafficIn
  • TrafficOut
  • IOPS
  • Storage capacity