SYSTEM
READINESS
Handling Volume
Much statistical information
from current systems plus
extensive
testing of the new
application architecture
made it evident to us that
there was a potential for
serious performance
issues if users were allowed
to run ad hoc queries and
reports against global
databases. Therefore, a
database instance
was developed that
duplicated the online
transaction-processing
(OLTP) instance for
reporting purposes. From
this
duplicated table structure,
ad hoc queries and standard
reports of any
configuration could be run
without affecting
the performance of daily
business transaction
processing.
Maintaining a 7x24x365
Application
With worldwide access to the
application 28 locations,
the system was being used
virtually all the time.
There was no quiet period.
Therefore, careful planning
had to
be done to minimize the
impact of system down time
for maintenance activities
on any one location.
Testing, Testing, and More
Testing...Are We
There Yet?
Developing and executing
application and interface
test
cases was probably one of
the most time-consuming portions
of our implementation from
an SA/BA perspective.
However, it was also one of
the most crucial from a
system readiness
perspective. Our SA/BA
resources developed
these cases and executed
them with additional
help from ETMs. These test
cases provided a wealth of
information necessary to
troubleshoot problems as
well
as give us a good feel for
the functionality was
performing
the way we thought it would
perform. By running
several rounds of testing,
we were effectively able to
attack
and resolve most issues with
system functionality
and performance.
Performance Under Simulated
Business
Conditions...A Day in the
Life
A technique we found most
useful in the final months
before cutover was the " day
in the life" test. This involved
having all BAs, ETMs, and
numerous others from the
business trained in the new
system sign on
and perform all the normal
transactions that they
would be required to do
during a routine workday.
This provided us an
excellent view of how the
system
would react under load and
where we could expect to
see peaks and valleys in the
daily processing cycle. We
included several
rounds of testing,
performance tuning
the system after each round
to increase system
efficiency.
Go or No Go?
Once all rehearsals and "day
in the life" tests were complete
and issues resolved, a last
round of go/no go meetings
were held with key
stakeholders worldwide to
address any issues of
concern. One key item here
was
that our champion and team
management staff all had
to be in total agreement
before the "all systems go"
sign-off was given. All team
members were then notified
and
schedules set.