Cisco Cisco Customer Voice Portal 8.0(1)

•

•

•

•

•

•

Data Retention

Via the Operations Console, users are able to select the time of day to run database purge, and
to set the number of days of data to be retained by data cate

gory. During schema creation, default

data retention values are specified for each data category. Note that a high level category, such
as Call, cannot have a lower retention time than a dependent category, such as Call Event.

When you schedule a purge from the Operations Console, two Windows jobs are scheduled

on the Reporting Server. The time that you choose is used to schedule what is referred to as the
"Nightly purge" job. The other job is referred to as the "Midday purge" and is automatically
scheduled 12 hours from the Nightly purge. So, for example, if you schedule a purge at 2 A.M.,
then the Nightly purge is run at 2 A.M. and the Midday purge at 2 P.M.

• The Nightly purge performs a purge if necessary (as required by a data retention value, or

for an emergency purge—see below), in addition to other tasks like updating the database
statistics. If a purge is performed, the statistics are updated after the purge. In addition, on
Sundays, the Nightly purge also copies the Informix log file to a backup directory, creates a
new log file and deletes the old. The Nightly purge should be scheduled at a time of low call
and reporting volume.

• The midday purge is the same as the nightly purge. In the event that data volume spikes

during the day and an emergency purge is required, it will be handled at midday. Midday
purge also serves as a backup for the nightly purge. If the nightly purge fails to allocate new
fragments for new data, this will be taken care of by the midday purge. Even if a purge occurs,
this process is not system intensive in the same way that the Nightly purge is.

• During a purge, reporting users may be disconnected from the database, if they are holding

locks that contend with purge. Notify reporting users not to run reports at this time.

If the number of days of data that you chose to retain cannot be contained within the database,
then the database will 'emergency' purge old data to create space for new data. Emergency purge
is, therefore, a critical safety mechanism for Unified CVP. If used space has grown past the
system's high water mark, the user will be notified by an SNMP trap message and data will be
purged. The SNMP notification will alert the user to the loss of data and request that the

y shrink

their 'retention days' data settings. Users should reduce the number of actual days of data retained
until emergency purge is no longer required. Additionally, users can reduce the data generated
by means of data filters (for VXML Server application detail data filtering, see 

During a database purge operation, the Reporting Server disconnects from the database (though
for no more than 10 minutes) and starts buffering messages in memory until the purge is done.

Reporting Guide for Cisco Unified Customer Voice Portal Release 7.0(2)