The SERVICETIME monitoring program allows the service times of individual devices to be measured in detail. A check is carried out to establish that the disk devices are “attached”. The service times are recorded by DCS (Dynamic Channel Subsystem).
Distinctions are drawn between the following service times (see also “glossary”, “service times definition“).
Service time | System | Definition |
DQT | SM2 | Time from the I/O chain to the subchannel start |
FPT | DCS | Time from the subchannel start to I/O start |
DDT | DCS | Physical positioning time |
DCT | DCS | Data transfer time |
RST | SM2 | Time from channel end to device end |
In certain circumstances (see the notes below), DCS cannot provide any detailed service times. The service time is subdivided into the waiting time before the device and the hardware service time. The hardware service time is referred to as the remaining service time.
The data of the monitoring program is only output to the SM2 output file.
Notes on VM2000
Under VM2000 the recording of detailed service times of DCS can be started by only one guest system. If a second guest system attempts to activate the function as well, a warning is issued. The monitoring program is started, but detailed service times of DCS are not delivered.
The VM2000 command SHOW-VM-RESOURCES INFORMATION=*STD/*ALL can be used to check whether SERVICETIME measurement is already active in a guest system. If it is, message VMS2035 is output.
DCS cannot deliver detailed service times for disks with indirect I/O.
Note on x86 servers
DCS does not deliver detailed service times on x86 servers. Only the times measured by SM2 itself (device queue time, remaining service time) can be recorded.
Notes on disk storage systems
Given an overall hit rate of 100%, the device disconnect time is zero.
The following applies to parallel access volumes (PAV):
Base and alias devices can be selected.
Each device defined is measured locally. In particular, no monitored data of an alias device is added to a base device.