The following short presentation of the LM procedure serves to create a better understanding of connections and limits itself to the aspects essential for it.
Management of BS2000 devices
At the start of the LM, the attached tape devices (also emulated tape devices) are detached by the BS2000 system that needs to be migrated (/DETACH-DEVICE).
This is why, it’s recommended to stop processing the tapes before the LM to avoid jobs being aborted and so as to not hinder the DETACH command.
After the relocation the tape devices remain detached and must be attached if needed (/ATTACH-DEVICE).
Disks emulated by the local SU/MU are also detached before the LM, which is why their usage should be stopped before the migration.
PAV alias devices on SU /390 are also automatically detached before the LM, but are attached again after the LM on the target SU.
Pay attentions to the hints in chapter -> Further notes about LM.
An LM in BS2000 consists of multiple phases that follow one another:
Phase 1 – Setup phase
In the setup phase the preconditions for the LM are checked (see chapter -> Usage scenarios and preconditions for LM). If all prerequisites are fulfilled then the necessary process environment is provided for the system under migration to the target SU (the VM is furnished with the necessary main memory and assigned devices on the target SU).
Hints:
- The necessary main memory is allocated on the target SU and this means it’s no longer free.
However in the system overviews of SEM, the inbound BS2000 system is not visible on the target SU.
This creates a hole in SEM to the end of the LM in the storage display – the free main memory is smaller than the sum of main memory values of the displayed BS2000 systems would make one expect.
In VM2000 the inbound BS2000 system is visible on the target SU. - The same is valid in SEM also for the devices assigned to the BS2000 system on the target SU: though they aren’t free, they are also not shown as assigned.
Phase 2 – Copying the main memory from the source SU to the target SU
In the first step all the pages of the main memory of the system to be migrated are transferred from source SU to target SU. Additionally, VM200/X2000 starts monitoring the working memory of the system to be migrated on the target SU and notes every subsequent change.
In the next step, the memory pages that have been changed in the meantime are transferred again.
This step is repeated multiple times depending on necessity, where usually the number of the pages to be copied becomes smaller with every run.
At the end of phase 2, the I/O system is stopped for the BS2000 system under migration. While the programs can start I/Os, they are not executed, but instead the I/O tasks are noted (by VM2000/X2000) and started only after migration on the target SU. This way the number of changing memory pages is diminished further.
At the very end of Phase 2, a last storage transfer is carried out.
The transfer of the main memory is done through the internal network -> MCNPR.
This phase of the LM lasts longest and depends on the size of the main memory, the transfer speed in the MCNPR and on the activity within the BS2000 system (how often the individual sites are changed during LM).
The BS2000 and the running applications are running in this phase without impairment, however, the CPUs are loaded additionally on SU /390 in monitor system by transferring the main memory.
Phase 3 – Transfer of the system to the target SU
If the whole memory (or at least a large part) is successfully transferred, then the system is stopped on the source SU and the possibly not transferred storage sites and current register content are copied to the target SU.
Afterwards, the already migrated system is continued on the target SU at the point when it was interrupted and the I/O processing is also restarted.
The time between stopping the system on the source SU and restarting it on the target SU is usually only a few seconds.
If restarting the system on the target SU fails, then the LM is aborted and the system is continued on the source SU. All preparations made on the target SU are reversed.
Phase 4 – Post-treatment
After restarting the system on the target SU, resources reserved for this system are released on the source SU (devices, main memory, etc.).
Important BS2000 components such BCAM, I/O control and MSCF are integrated into the LM to avoid automatic MSCF error recognition and to adjust the own I/O configuration on the target SU.