PVS - dynamic Disk Management System
detailed Informations
The Present Situation while Processing Sequential Datasets
Due to the large dasd capacity, available nowadays, is seems to
be usefull, to store sequential datasets on dasds und to hold
multiple generations of a dataset. Under VSE resp. z/VSE
this is only possible with a substantial effort, as
sequential dasd datasets need an EXTENT statement with the
exact specification of start track and size of the dataset
as well as an ASSGN statement, pointing to the appropriate
device
the label in the DLBL statement must be changed, if
multiple generations of a dataset are to be retained.
A solution for these problems offers PVS - the dynamic disk
management system.
Structure and function of PVS
The user defines, which dasds or dasd areas are to be controlled
by PVS. These areas are called pools. Any number of pools may
be specified, each with up to 16 areas. A pool can be accessed
by a single or multiple real or virtual cpus concurrently.
PVS owns a unique catalog, in which all controlled, sequential
datasets are entered. The datasets may be controlled by
expiration date and/or number of generations.
PVS must be initialized during IPL in any partition. After
initialisation, the partition is free again. PVS uses the
vendor exits from the operating system to become active.
Thus PVS gets control while a dasd file is opened or closed,
if "end of extent" occurs and while end of task processing is
performed.
Output of Sequential Dasd Datasets
Whether a sequnial output file is to be managed by PVS is
specified by a specification of a relative start track of one
in the EXTENT statement. Besides, PVS becomes active, if no
EXTENT Statement is specified.
PVS needs a POOL Name for the allocation of the area for the output
dasd file. Optionally, the pool name can be specified globally or
explicitly in the EXTENT statement. In the latter case, the pool
name is written in the place of the volser specification in the
EXTENT statement.
PVS manages the usage of the pool areas by itself. Therefore, it
can add the missing start track and size to an EXTENT statement
or can add an EXTENT statement if none was specified to the
DLBL statment. Thus, the annoying changing of these JCL
statements can be omitted.
Optionally, an output dataset can be handled as a "single file"
- that means, there is always onle one edition of this
dataset - or as a "generation file" - which means, any number
of editions of a dataset may be retained (under z/OS this
is handled as a "generation-data-group" / GDG).
In the case of a GDG output, PVS supplements the file-id
(44 characters in length) in its last 4 to 6 positiions by
the generation numer in the form "G#" followed by a two to
four numeric number indicating the generation of the dataset
(G#01 to G#9999).
Each new output of such a dataset increments the generation
number by one. Each dataset holds the generation number
assigned to it as long as it exists. The highest possible
genration number is 9999.
Each dataset written to a pool is managed by PVS dynamically,
that means, without any preceeding definition in the PVS
catalog.
Optionalle, additional specifications may be made for a dataset.
Either system wide or specificalle to this dataset.
Those specifications may be the size of the first extent,
(primary extent) the size and the number of the sequence
extents as well as the expriation date and the nbr of generations
to be retained.
When closing an output dataset, PVS releases the excessive dasd
space, not needed for the file. This area may be immediately
used by another file.
Only when a "normal" end of task is reache, all datasets created
by this task are permanently entered to the PVS catalog. If there
is an abnormal end of task, even the already closed datasets
are deleted and the dasd space is released.
Input of a Sequential Dasd Dataset
With the open of a sequentila input dataset PVS becomes
unconditionally active.
PVS checks, if the input dataset, only specified by a DLBL
statement, is registered in its catalog. If yes, PVS
completes the DLBL statement by one or more EXTENT statements.
If necessary, an assignment is also build to the appropriate
dasds.
If the input dasd has multiple generations (GDG), PVS selects
usually the latest generation. Ofcourse, the user can specify
a specific (generation nnnn) or relative (latest generation -n)
generation in the DLBL statement.
Backup of the User Datas
PVS owns a flxible tool to backup the user data stored in one
or more pools. The backup medium is always a tape or cartridge.
Due to the high capacity of 30 GB and more, a restore of even
a small file may take substantial time, as the tape must be
read sequentilly to positition it to this file.
For this reason, PVS creates an index file while backing up
the user data, which holds the physical position of each
file on the tape. This file may be stored as a dasd dataset
or as a sequence file on the backup tape or even both.
In the case of a restore, PVS can position the backup tape
quickly to the required position to perform the desired
restore.
Advantages of PVS
simple and flexible administration of sequential dasd datasets
as no EXTENT statements are necessary
any number of generations of a dataset may be retains without
any manuall effort .
unused dasd space within a file is released at close
dasd datasets need not be specified in advance. Just
creating the datasets in a pool is enough, that PVS
adds it to its catalog
fast restore of a PVS controlled dataspace due to the
possibility of hardware initiaded positioning of the
restore media.
Other
Since PVS is a new development, the hardware and software
standards valid today could be respected.
PVS does not support ISAM and direct access datasets, as those
are mostly converted by the operating system to VSAM datasets.
As nowadays dasd units usually own an own cache storage, PVS
uses the regular VTOC without any indexing. Thus the overhead
for I/Os and cpu time is avoided.
PVS has a transfer program to convert the catalog of other
disk management systems into a PVS catalog.
Systemintegrity in PVS
PVS uses the vendor exits of VSE. Thus only official interfaces
of the operating system are used.
Maintenance and Development of PVS
PVS is developed and maintained in Germany. The manuals
are available in German and English language.
verfügbar.
click here to go back to the general informations:
to download the flyer of PVS in PDF format point to
the icon, klick right mouse button and select "store
destination";
to display the flyer click left mouse button