Configuring LIBNAME Domain Disk Space

The primary file system is the base directory that you assign to the LIBNAME domain by issuing a PATHNAME= statement in the SPD Server LIBNAME parameter file, libnames.parm. Here is an example of a libnames.parm parameter file entry:

   spdsserv -acl
      -acldir InstallDir/site
      -nameserver samson
      -libnamefile libnames.parm

Sample libnames.parm file entry for a UNIX system:

LIBNAME=all_users 
  pathname=/disk1/peruser_tables;

Sample libnames.parm file entry for Windows:

LIBNAME=all_users 
  pathname=d:\peruser_tables;

When SPD Server users create new tables in a LIBNAME domain, you must keep in mind that thee metadata component (.mdf) must start in the primary file system. If all the available space in the primary file system is consumed, SPD Server cannot create new tables until disk space becomes available.

Example 1 stores all the component files (metadata, data, and index data) in the primary file system. This arrangement can cause problems if you use large tables. Large tables can quickly fill up the primary file system. To avoid this problem, you should store the data and index components separately from the primary file system. Example 2 shows how to do this using ROPTIONS= with your LIBNAME statement in your libnames.parm file.

The following SPD Server code invokes the libnames.parm file:

   spdsserv -acl
      -acldir InstallDir/site
      -nameserver samson
      -libnamefile libnames.parm

Sample libnames.parm for a UNIX system:

   LIBNAME=all_users pathname=/disk1/peruser_tables
      roptions="datapath=('/disk2/userdata' '/disk3/userdata'
                          '/disk12/userdata' '/disk13/userdata')
      indexpath=('/disk4/userindexes' '/disk5/userindexes'
                 '/disk14/userindexes' '/disk15/userindexes')";

Sample libnames.parm for a Windows system:

   LIBNAME=all_users pathname=d:\peruser_tables
      roptions="datapath=('e:\userdata' 'f:\userdata')
         indexpath=('g:\userindexes' 'h:\userindexes')";

In Example 2, the PATHNAME= directory stores metadata files for SPD Server tables in the all_users LIBNAME domain. The initial and overflow stores for the data and index files are directed to other file systems. In Example 2, users who create large tables will not quickly exhaust the primary file system; the primary file system is reserved for only small metadata files. The larger data and index files are stored in the other file systems specified with the DATAPATH= and INDEXPATH= options in the LIBNAME parameter file.

If you need to add file systems to your path because an existing file system is running out of space, see the following:

   spdsserv -acl
      -acldir InstallDir/site
      -nameserver samson
      -libnamefile libnames.parm

Sample libnames.parm for a UNIX system:

LIBNAME=all_users pathname=/disk1/peruser_tables
   roptions="datapath=('/disk2/userdata' '/disk3/userdata'
                       '/disk12/userdata' '/disk13/userdata')
   indexpath=('/disk4/userindexes' '/disk5/userindexes'
              '/disk14/userindexes' '/disk15/userindexes')";

Sample libnames.parm for a Windows system:

LIBNAME=all_users
pathname=d:\peruser_tables
   roptions="datapath=('e:\userdata'
                       'f:\userdata'
                       'i:\userdata')
   indexpath=('g:\userindexes'
              'h:\userindexes'
              'j:\userindexes')";

In Example 3, SAS users can continue to create more SPD Server tables, as long as space is available for the metadata files in the primary file system. When the primary file system is exhausted, you must create a new LIBNAME domain. You cannot expand storage for the .mdf components by adding the METAPATH= specification to your ROPTIONS= value in your LIBNAME parameter file. Remember the SPD Server restriction : the initial partition file for all .mdf components must be created in the primary file system (the directory that was first specified by the PATHNAME= option for the LIBNAME domain).