Creating Analysis Services Cubes - SQL Server Analysis Services 2012 Cube Development - page 110

Merging partitions
Merging multiple partitions is common as the partitions age (become less frequently
queried) or when you need to reduce the total number of database files. For example,
in the year 2013, we might not be particularly interested in data from the 1990s but
could expect historical data to be queried occasionally. If so, we could keep monthly
partitions for the current year but merge partitions for each previous year into yearly
partitions and perhaps even merge all of the 1990s data into a decade-level partition.
If you work in an environment where data must be aggregated many times each day,
you might create multiple partitions throughout the day and later merge all intraday
partitions into a single daily partition. As you will learn in Chapter 8 , Administering
and Monitoring Analysis Services , synchronization speed is largely dependent on the
number of files that must be transferred. Therefore, merging partitions could also op-
timize synchronization speed.
How to do it...
To merge partitions you must identify the source and target partitions. You can
use an empty target partition or an existing partition that already has data. In either
case both source and target partitions must be fully processed before you can merge
them. Suppose we would like to merge 2007 and 2006 partitions into the main par-
tition called Fact Reseller Sales , which will act as the target; for reference, perform
the following steps:
1. Right-click on the Partitions folder within SSMS, and select Merge Partitions
to open the Merge Partition dialog: