56 Storage Management with DB2 for OS/390

6.1.8 TableSpace and Index Space Names for SMS

The recommendation in this book forfinely tuned SMS installations is to imbed
SMS codes into the namesof DB2table and index spaces. This is shown in Table
16 on page 56. The data set names have the structure shown inTable 3 on page
23,with a change in the space name itself. As explained in 6.1.7, “AssigningS MS
Classes to DB2 TableSpaces and Index Spaces” on page 53, this name contains
codes for the ACS routines.The ACS routines use these codes to establish
Storage Classes, ManagementClass es and Storage Groups.
Table16. TableSpace and Index Space Names with SMS Codes
The elements oft hespace name are:
uvssssss Space name:
u Storage Class code
v Management Class code
ssssss User assigned name

6.1.9 Managing PartitionedTable Spaces with SMS

With DB2, the user has choices on how to allocate and distribute the individual
data sets ofa par titionedtable space. Two examples areshown in:
• Appendix A, Section A.2, “Partitioned TableS pace,DB2 Defined, Without
SMS” on page 162
• Appendix A, Section A.3, “Partitioned TableS pace,User Defined, Without
SMS” on page 164
SMS can also beused to distribute the individual partitions. Several different
methods are possible; forexample:
• Let SMS manage everything.
• Use one SMS Storage Group for each partition.
• Use one SMS Storage Group for one partitioned table space.
Let SMS Manage Everythi ng
For many partitioned tablespaces and index spaces, SMS can handle the
partition distribution. If the number of volumes in the Storage Group is much
largerthan the number of par titions,then SMS willmost likely place each partition
on a separate volume.
Storage group SGDB20 is an example of this case. Tablespaces and index
spaces are allocated byS MS on these volumes,try ingt o balance theI/O activity
on the volumes. This method should be adequate for manyinstallations.This
method is the preferred techniquefor storage administrators, because it has the
advantage of simplicity.
This method can be used for table spaces and partitions, where each data set is
a large fraction of a volume.Because a volume can only handle one partition,the
separation is automatic. On the other hand, space fragmentationon thevolumes
hlq.DSNDBx.dbname.uvssssss.ynnnn.Ammm