v7000 MDisks Test

A managed disk (MDisk) is a logical unit of physical storage. MDisks are either arrays (RAID) from internal storage or volumes from external storage systems. MDisks are not visible to host systems.

An MDisk might consist of multiple physical disks that are presented as a single logical disk to the storage area network (SAN). Each MDisk is divided into a number of extents, which are numbered, from 0, sequentially from the start to the end of the MDisk.

Real capacity to the VDisks/volumes is provided by groups of MDisks, known as storage pools. This is because, volumes are created from the extents that are available in these pools. It is hence evident that I/O requests to a volume are in effect serviced by the MDisks in the pools from which that volume was created. This means that a processing bottleneck in an MDisk can significantly delay reading from and writing to the VDisk that overlays it, and can thus adversely impact the overall I/O performance of the storage system. This is why, it is important that administrators track the I/O load on and the I/O processing speed of each MDisk in each node of the IBM Storwize v7000 system. The v7000 MDisks test helps administrators achieve this. This test monitors the I/O operations on each MDisk in each node of the storage system, reports how well every MDisk processes the I/O, and reveals processing latencies (if any) in that MDisk. This way, MDisks that are overloaded, those that are slow in processing I/O, and the nodes that such MDisks are mapped to, can be accurately identified.

Target of the test : An IBM Storwize v7000 storage system

Agent deploying the test : A remote agent

Outputs of the test : One set of results for every MDisk in each node of the storage system being monitored.

Configurable parameters for the test
Parameter	Description
Test Period	How often should the test be executed.
Host	The host for which the test is to be configured.
Port	The port number at which the specified host listens to. By default, this is NULL.
Timeout	Specify the duration (in seconds) beyond which the test will timeout in the Timeout text box. The default value is 60 seconds.

Measurements made by the test
Measurement	Description	Measurement Unit	Interpretation
Read operations	Indicates the rate at which read operations were performed by this MDisk of this node.	Reads/Sec	Comparing the value of each of these measures across MDisks helps you in identifying the overloaded MDisks - it could shed light on irregularities on load balancing across the MDisks.
Write operations	Indicates the rate at which write operations were performed by this MDisk of this node.	Writes/Sec
Read data	Indicates the rate at which data blocks were read from this MDisk of this node.	MB/Sec	By comparing the value of each of these measures across MDisks, you can identify the MDisk that is the slowest in reading and writing. The reason for the slowness has to be determined and eliminated to ensure the high availability and performance of the storage system.
Write data	Indicates the rate at which the data blocks were written to this MDisk of this node.	MB/Sec
Average read external response	Indicates the average time taken by this MDisk to respond to read requests.	Millisec	For each SCSI Read command sent to the Managed disk, a timer is started when the command is issued across the fibre channel from the host system to the IBM Storwize v7000 storage system. When the command completes, the elapsed time since the timer was started is added to the cumulative counter. A low value is desired for this counter. A high value is indicative of slowness when responding to read requests. The least responsive MDisk can be identified by comparing the value of this measure across MDisks.
Average write external response	Indicates the average time taken by this MDisk to respond to write requests.	Millisec	A low value is desired for this counter. A high value is indicative of slowness when responding to write requests. To know which MDisk is taking an unreasonably long time to service write requests, compare the value of this measure across MDisks.
Peak read external response	Indicates the maximum time taken by this MDisk to respond to a read request during the last measurement period.	Millisec	Compare the value of this measure across the MDisks to identify the MDisk that was the slowest in responding to read requests during the last measurement period. If the same MDisk tops this comparison consistently, it could indicate a read I/O processing bottleneck in that MDisk.
Peak write external response	Indicates the maximum time taken by this MDisk to respond to a write request during the last measurement period.	Millisec	Compare the value of this measure across the MDisks to identify the MDisk that was the slowest in responding to write requests during the last measurement period. If the same MDisk tops this comparison consistently, it could indicate a write I/O processing bottleneck in that MDisk.
Peak read queued response	Indicates the maximum time a read request was waiting in the queue before being sent to this MDisk during the last measurement period.	Millisec	The value of this measure includes the time spent by the read request in the queue and the time taken for execution of this request by the MDisk. By comparing the value of this measure with that of the Peak read external response measure, you can understand where the read requests could have spent maximum time – in the queue? or in the MDisk, being processed?
Peak write queued response	Indicates the maximum time a write request was waiting in the queue before being sent to this MDisk during the last measurement period.	Millisec	The value of this measure includes the time spent by the write request in the queue and the time taken for execution of this request by the storage controller. By comparing the value of this measure with that of the Peak write external response measure, you can understand where the read requests could have spent maximum time – in the queue? or in the MDisk, being processed?