Performance Indicators for HDFS

Table 1 Performance indicators for HDFS

Item

Definition

Significance

Cluster

Avg. CPU Usage (%)

CPU usage in a statistical period. By default, one statistical period is ten seconds.

Indicates the service load on the CPU.

Avg. Memory Usage (%)

Memory usage in a statistical period. By default, one statistical period is ten seconds.

Indicates the service load on the memory.

Max. CPU Usage (%)a

Maximum CPU usage of a statistic object in a sampling period

Indicates the maximum CPU utilization of all nodes.

Transmit Bandwidth (KB/s)ab

Data volume sent by all cluster ports in a sampling period

Indicates the data volume sent by a cluster in a sampling period.

Receive Bandwidth (KB/s)ab

Data volume received by all cluster ports in a sampling period

Indicates the data volume received by a cluster in a sampling period.

FTP Upload Bandwidth (KB/s)

Average data volume of FTP upload requests processed by a module per second

Indicates the FTP upload data volume per second.

FTP Download Bandwidth (KB/s)

Average data volume of FTP download requests processed by a module per second

Indicates the FTP download data volume per second.

Read Bandwidth (KB/s)b

Average read request data volume processed by a module per second

Indicates the data volume read per second.

Write Bandwidth (KB/s)b

Average write request data volume processed by a module per second

Indicates the data volume written per second.

Bandwidth (KB/s)b

Average I/O data volume on a module per second

Indicates the I/O data volume per second.

Read IOPS (IO/s)b

Average number of read requests processed by a module per second

Indicates the number of read I/O requests received per second.

Write IOPS (IO/s)b

Average number of write requests processed by a module per second

Indicates the number of write I/O requests received per second.

Total IOPS (IO/s)b

Average number of I/O requests processed by a module per second

Indicates the number of I/O requests received per second.

HDFS Concurrenciesb

Number of HDFS concurrencies on a statistic object

Indicates the number of HDFS concurrencies on clusters or nodes.

Node

Avg. CPU Usage (%)

CPU usage in a statistical period. By default, one statistical period is ten seconds.

Indicates the service load on the CPU.

Avg. Memory Usage (%)

Memory usage in a statistical period. By default, one statistical period is ten seconds.

Indicates the service load on the memory.

FTP Connections

Number of connections between a statistic object and FTP clients

Indicates the number of connections between nodes and FTP clients.

FTP Upload Bandwidth (KB/s)

Average data volume of FTP upload requests processed by a module per second

Indicates the FTP upload data volume per second.

FTP Download Bandwidth (KB/s)

Average data volume of FTP download requests processed by a module per second

Indicates the FTP download data volume per second.

Read Bandwidth (KB/s)b

Average read request data volume processed by a module per second

Indicates the data volume read per second.

Write Bandwidth (KB/s)b

Average write request data volume processed by a module per second

Indicates the data volume written per second.

Bandwidth (KB/s)b

Average I/O data volume on a module per second

Indicates the I/O data volume per second.

Read IOPS (IO/s)b

Average number of read requests processed by a module per second

Indicates the number of read I/O requests received per second.

Write IOPS (IO/s)b

Average number of write requests processed by a module per second

Indicates the number of write I/O requests received per second.

Total IOPS (IO/s)b

Average number of I/O requests processed by a module per second

Indicates the number of I/O requests received per second.

HDFS Concurrenciesb

Number of HDFS concurrencies on a statistic object

Indicates the number of HDFS concurrencies on clusters or nodes.

Heterogeneous device tiering performance

Read Bandwidth (KB/s)

Average data volume of data retrieval requests processed by a module per second

Indicates the volume of data retrieved from a heterogeneous device per second.

Write Bandwidth (KB/s)

Average data volume of data migration requests processed by a module per second

Indicates the volume of data migrated to a heterogeneous device per second.

Bandwidth (KB/s)

Average I/O data volume on a module per second

Indicates the I/O data volume per second.

Read IOPS (IO/s)

Average number of data retrieval requests processed by a module per second

Indicates the number of I/Os retrieved from a heterogeneous device per second.

Write IOPS (IO/s)

Average number of data migration requests processed by a module per second

Indicates the number of I/Os migrated to a heterogeneous device per second.

Total IOPS (IO/s)

Average number of I/O requests processed by a module per second

Indicates the number of I/Os per second.

Remote replication pair

Synchronization Duration (s)a

Synchronization duration in the statistical period

Indicates the distribution of synchronization duration (s) in asynchronous remote replication.

Time Since Last Synchronization (s)ab

Time interval between the current system time and the start time of last remote replication

Indicates the time interval between the current system time and the start time of last remote replication

Logical Bandwidth (KB/s)a

Average I/O data volume on a module per second before compression

Indicates the I/O data volume per second before compression.

Time Difference (s)ab

Time difference between the data synchronization points in time of the primary and secondary resources in a synchronization period

Indicates the time difference between the data synchronization points in time of the primary and secondary resources in a synchronization period of asynchronous remote replication.

Storage pool

Capacity Usage (%)a

Percentage of the used capacity of a storage pool

Indicates the percentage of the used capacity in a storage pool.

Disk

Read Bandwidth (KB/s)b

Average read request data volume processed by a module per second

Indicates the data volume read per second.

Write Bandwidth (KB/s)b

Average write request data volume processed by a module per second

Indicates the data volume written per second.

Total Bandwidth (KB/s)b

Average I/O data volume on a module per second

Indicates the I/O data volume per second.

Read IOPS (IO/s)b

Average number of read requests processed by a module per second

Indicates the number of read I/O requests received per second.

Write IOPS (IO/s)b

Average number of write requests processed by a module per second

Indicates the number of write I/O requests received per second.

IOPS (IO/s)b

Average number of I/O requests processed by a module per second

Indicates the number of I/O requests received per second.

Avg. Operation Latency (ms)b

Average latency of read and write operations on a statistic object in a sampling period

Indicates the time required to process I/Os.

Avg. Write Operation Latency (ms)b

Average latency of write operations on a statistic object in a sampling period

Indicates the time required to process write I/Os.

Avg. Read Operation Latency (ms)b

Average latency of read operations on a statistic object in a sampling period

Indicates the time required to process read I/Os.

Avg. Operation Latency (us)

Average latency of read and write operations on a statistic object in a sampling period

NOTE:

Compared with Avg. Operation Latency (ms), Avg. Operation Latency (us) is more accurate and the monitored performance data is more accurate and complete.

Indicates the time required to process I/Os.

Avg. Write Operation Latency (us)

Average latency of write operations on a statistic object in a sampling period

NOTE:

Compared with Avg. Write Operation Latency (ms), Avg. Write Operation Latency (us) is more accurate and the monitored performance data is more accurate and complete.

Indicates the time required to process write I/Os.

Avg. Read Operation Latency (us)

Average latency of read operations on a statistic object in a sampling period

NOTE:

Compared with Avg. Read Operation Latency (ms), Avg. Read Operation Latency (us) is more accurate and the monitored performance data is more accurate and complete.

Indicates the time required to process read I/Os.

Usage (%)b

Request processing time and queuing time/Statistical period

Indicates the usage of the specified object.

Avg. Read I/O Size (KB)

Average read I/O size from the last sampling time point till now

Indicates changes in the read I/O size.

Avg. Write I/O Size (KB)

Average write I/O size from the last sampling time point till now

Indicates changes in the write I/O size.

Port

Read Bandwidth (KB/s)b

Average read request data volume processed by a module per second

Indicates the data volume read per second.

Write Bandwidth (KB/s)b

Average write request data volume processed by a module per second

Indicates the data volume written per second.

Bandwidth (KB/s)b

Average I/O data volume on a module per second

Indicates the I/O data volume per second.

Read IOPS (IO/s)

Average number of read requests processed by a module per second

Indicates the number of read I/O requests received per second.

Write IOPS (IO/s)

Average number of write requests processed by a module per second

Indicates the number of write I/O requests received per second.

IOPS (IO/s)

Average number of I/O requests processed by a module per second

Indicates the number of I/O requests received per second.

Replication link

IOPS (IO/s)a

Average number of I/O requests processed by a module per second

Indicates the number of I/O requests received per second.

Sending Bandwidth for Replication (KB/s)a

Average data volume that is sent from the local device to a remote device per second

Indicates the I/O data volume sent per second.

Receiving Bandwidth for Replication (KB/s)a

Average data volume that is received from a remote device to the local device per second

Indicates the I/O data volume received per second.

Avg. I/O Response Time (ns)a

Time of processing one request by a module, including the waiting time

Indicates the speed of I/O request processing. The longer the queue is, the longer the system takes to respond to a request. For a specified load, response time is in direct proportion to queue length.

Max. I/O Response Time (ns)a

Maximum average I/O response time in all statistical periods after the performance statistics collection starts. The average I/O response time refers to the time required to process one read or write request within a statistical period. The response time includes the waiting time. The default statistical period is five seconds.

Indicates the maximum duration of the system responding to a read/write request.

a: Indicates that the performance indicator does not support the Grafana service.

b: Indicates that monitoring thresholds can be configured for the performance indicator. For details, see Managing Monitoring Thresholds.