The information shown is based on MONITOR domain and I/O domain SAMPLE data.
The first entry is reserved for averages since the last RESET subcommand or since data collection was activated.
It is designated by the >>Mean>> entry in the Interval End Time column.
Field Descriptions:
Interval End Time
The end time of the monitor interval for which the values were calculated.
In the case of the >>Mean>>entry, you can find the relevant interval in the
header line.
Volume Label
The volume label.
For the >>Mean>> row or for a volume-group row, this column contains the number of volumes.
WA
Indicates where the volume's base RDEV is attached:
CP
Indicates that it is in the CP-owned list.
SY
Indicates that it is attached to SYSTEM.
GU
Indicates that it is attached to a guest.
If the volume is not attached, this field contains dots.
LCU SSID
The subsystem ID (SSID) of the logical control unit (LCU).
T
Indicates the RDEV type for the volume's base RDEV:
P
For a PAV base
H
For a HyperPAV base.
If it is neither a PAV base nor a HyperPAV base, this field contains a dot.
Mdisk Links
The number of minidisk links to the volume.
I/O
The I/O rate. This is the number of SSCH and RSCH operations executed per second to this
volume through the base RDEV or any alias RDEV working on behalf of the
base during the selected interval.
AIOR
The alias I/O ratio. For each I/O operation performed to the base RDEV,
this is the number of I/O operations performed through the alias RDEVs.
Volume Times (msec):
Pend
The "function pending" time.
This is the average amount of time that an I/O request remained pending in the I/O subsystem due to path-busy conditions.
This value and the Disc, Conn, Serv, Busy, and Resp values
are calculated based on information obtained from the
subchannel measurement blocks.
Dots are inserted for volumes where the information is not available or is not calculated.
Disc
The "volume disconnected" time. This is the average amount of time that
the volume remained disconnected from the channel while executing I/O
requests. It
includes the time spent on executing the SEEK
and SET SECTOR orders and also any re-connect misses. High values could
indicate overloaded paths or many long SEEKs to data at opposite ends of
the volume.
The value also includes the "device-active-only" time where that
information is available.
Conn
The "volume connected" time. This is the average time that the
volume was connected to the channel while executing I/O requests,
that is, receiving commands or actually transferring
data. It
also includes the SEARCH time.
Serv
The service time for the volume. This is the sum of the
function pending, connected, and disconnected times.
Wait
The volume wait time.
Resp
The volume response time.
This is the sum of the service time and the wait time.
CUWt
The control unit queuing time, that is, the additional wait time caused
when the target device was already busy with an I/O operation that was
started from another system.
Req. Qued
The average I/O queue length in the selected interval.
The value is taken from CP monitor high-frequency sample data;
it is an indicator of I/O contention on the volume.
The I/O queue length
(Req. Qued) is
not meaningful for volumes
that are dedicated (attached) to guest systems. The guest system will
not try to start another I/O operation before an active one has ended,
but rather will do its own queuing. Guest system queuing cannot be shown
by host performance monitors, so they will have to be analyzed using a
performance monitor on the guest system.
The queue length reported here is the number of real I/O operations (channel programs)
that CP's real I/O layer is waiting to start to the device or volume.
This includes I/O operations performed for all reasons: for minidisk I/O, for paging I/O, and for spooling I/O, for example.
For devices in the CP-owned list, additional queueing information can be found on one of these screens:
The percentage of time during which the volume was busy.
This value could be greater than 100 if there are alias RDEVs
doing work on behalf of the base RDEV.
Serv:
Base
The average amount of service time per I/O operation through the base RDEV to the volume.
Alias
The average amount of service time per I/O operation through an alias RDEV to the volume.