Many
companies finish consolidating a data center by announcing that they have now
integrated all their zOS partitions into one system. No thought is given to
additional cost-saving opportunities that result from this integration.
Sparkassen
Informatik GmbH, data center operator of more than half of all German Sparkassen
(savings and loan associations), has gone a step further.
The
company is the result of the merger of several operating companies and inherited
four different dataset naming systems. In daily operation these differences led
to a lot of extra time spent on maintenance. In addition, introducing new
applications was made unnecessarily difficult. That’s why Sparkassen
Informatik decided to introduce a new, consistent dataset naming system.
Four
productive SYSPLEX systems, a validation LPAR and a test LPAR had to be
integrated .
The
IT departments of the different banks use identical programs, but the dataset
names they use are different. The IT departments of the various banks use
identical programs, but their dataset names are different. These dataset names
are made up of variables that result from JCL procedures. These procedures and
variables had to be taken into account for every single dataset in order to
ensure consistent conversion.
A
further requirement was that the conversion had to take place on one weekend and
online operation was to be interrupted for as short a time as possible. No tapes
were to be converted, only datasets on DASD. Migrated datasets were also
excluded from conversion. However, they could be recovered manually if necessary
and assigned their new name. .
Sparkassen
Informatik opted for DSN/Change, a HORIZONT product. DSN/Change supports the
conversion of dataset names and all their dependent constructs in JCL, IDCAMS
and REXX. Conversion is consistent and can be cancelled and restarted at any
time. DSN/Change recognizes the link between symbolic and physical dataset
names. A large numbers of datasets can be converted in batch mode in a
relatively short amount of time. All changes are logged for recovery purposes
and can be simulated in advance.
Conversion
began in the test LPAR. A test environment is usually more problematic than a
productive environment for a conversion. In this case, though, conditions in the
test environment were similar to those in the productive environment. The actual
conversion took place on a weekend. 50,000 datasets with about 150,000
references in JCL, REXX and parameter datasets were converted. Issues arose due
to oversights, but these issues were brought under control within a day. The
knowledge gained from this was used by HORIZONT and Sparkassen Informatik to
improve the conversion process and refine the programs.
Next,
the validation LPAR was converted. This went a lot more smoothly than the first
conversion.
Dataset
conversion runs parallel to online operation. Datasets being used by online
programs are not converted. After the online programs are stopped, the datasets
being used by them are renamed and the programs are then restarted.
Three
productive data centers have now been converted. The number of datasets
converted ranged from 56,000 to 64,000 and the number of references from 179,000
to 246,000. These dataset names were referenced by up to 370 libraries. The time
needed for conversion was 22 hours per data center and online operation was
interrupted for a maximum of five hours. Sequential datasets, GDGs, VSAM
datasets, libraries, load libraries, job libraries and procedure libraries were
converted.
Sparkassen
Informatik now has client-specific, consistent dataset names. Administration has
been streamlined and new handover procedures now use standardized [JB1]dataset
names. As a result, costs have been reduced dramatically.
Many
companies finish consolidating a data center by announcing that they have now
integrated all their zOS partitions into one system. No thought is given to
additional cost-saving opportunities that result from this integration.
Sparkassen
Informatik GmbH, data center operator of more than half of all German Sparkassen
(savings and loan associations), has gone a step further.
The
company is the result of the merger of several operating companies and inherited
four different dataset naming systems. In daily operation these differences led
to a lot of extra time spent on maintenance. In addition, introducing new
applications was made unnecessarily difficult. That’s why Sparkassen
Informatik decided to introduce a new, consistent dataset naming system.
Four
productive SYSPLEX systems, a validation LPAR and a test LPAR had to be
integrated .
The
IT departments of the different banks use identical programs, but the dataset
names they use are different. The IT departments of the various banks use
identical programs, but their dataset names are different. These dataset names
are made up of variables that result from JCL procedures. These procedures and
variables had to be taken into account for every single dataset in order to
ensure consistent conversion.
A
further requirement was that the conversion had to take place on one weekend and
online operation was to be interrupted for as short a time as possible. No tapes
were to be converted, only datasets on DASD. Migrated datasets were also
excluded from conversion. However, they could be recovered manually if necessary
and assigned their new name. .
Sparkassen
Informatik opted for DSN/Change, a HORIZONT product. DSN/Change supports the
conversion of dataset names and all their dependent constructs in JCL, IDCAMS
and REXX. Conversion is consistent and can be cancelled and restarted at any
time. DSN/Change recognizes the link between symbolic and physical dataset
names. A large numbers of datasets can be converted in batch mode in a
relatively short amount of time. All changes are logged for recovery purposes
and can be simulated in advance.
Conversion
began in the test LPAR. A test environment is usually more problematic than a
productive environment for a conversion. In this case, though, conditions in the
test environment were similar to those in the productive environment. The actual
conversion took place on a weekend. 50,000 datasets with about 150,000
references in JCL, REXX and parameter datasets were converted. Issues arose due
to oversights, but these issues were brought under control within a day. The
knowledge gained from this was used by HORIZONT and Sparkassen Informatik to
improve the conversion process and refine the programs.
Next,
the validation LPAR was converted. This went a lot more smoothly than the first
conversion.
Dataset
conversion runs parallel to online operation. Datasets being used by online
programs are not converted. After the online programs are stopped, the datasets
being used by them are renamed and the programs are then restarted.
Three
productive data centers have now been converted. The number of datasets
converted ranged from 56,000 to 64,000 and the number of references from 179,000
to 246,000. These dataset names were referenced by up to 370 libraries. The time
needed for conversion was 22 hours per data center and online operation was
interrupted for a maximum of five hours. Sequential datasets, GDGs, VSAM
datasets, libraries, load libraries, job libraries and procedure libraries were
converted.
Sparkassen
Informatik now has client-specific, consistent dataset names. Administration has
been streamlined and new handover procedures now use standardized [JB1]dataset
names. As a result, costs have been reduced dramatically.
Gero
Saxlehner, project leader at Sparkassen Informatik: “Give me a data center and
I’ll convert it tomorrow afternoon.“ For questions: gero_saxlehner@informatik-kooperation.de
Josef Böck, responsible software developer at HORIZONT: “I’ve never
converted such an incredible number of datasets with a better project leader!“
Uwe Hahm, Managing Director of HORIZONT: “If we had known that, the price
would have been higher.”
Sabine Saxlehner, wife of the project leader: “Another weekend like that and
I’m moving out!”