Use the command-line utility mkstore to create and manage wallets.

Setting up a hierarchical file system store requires creating, registering, and mounting the store.

1. Call createStore.

   DBMS_DBFS_HS.createStore( store_name, store_type, tbl_name, tbs_name, cache_size, lob_cache_quota, optimal_tarball_size, schema_name);
   

2. Set mandatory and optional properties using the following interface:

   DBMS_DBFS_HS.setStoreProperty(StoreName, PropertyName, PropertyValue);
   

   For store_type = STORETYPE_TAPE, mandatory properties are:

   PROPNAME_DEVICELIBRARY, PROPNAME_MEDIAPOOL, PROPNAME_CACHESIZE.
    
   PROPNAME_CACHESIZE is already set by createStore. 
   

   You can change the value of PROPNAME_CACHESIZE using reconfigCache.

   Optional properties are:

   PROPNAME_OPTTARBALLSIZE, PROPNAME_READCHUNKSIZE, PROPNAME_WRITECHUNKSIZE, PROPNAME_STREAMABLE.
   

   For store_type = STORETYPE_AMAZONS3 mandatory properties are:

   PROPNAME_DEVICELIBRARY, PROPNAME_CACHESIZE, PROPNAME_S3HOST,PROPNAME_BUCKET, PROPNAME_LICENSEID, PROPNAME_WALLET.
   

   PROPNAME_CACHESIZE is set by createStore. You can change the value of PROPNAME_CACHESIZE using reconfigCache.

   Optional properties are:

   PROPNAME_OPTTARBALLSIZE, PROPNAME_READCHUNKSIZE, PROPNAME_WRITECHUNKSIZE, PROPNAME_STREAMABLE, PROPNAME_HTTPPROXY.
   

3. Register the store with DBFS Content API using:

   DBMS_DBFS_CONTENT.registerStore(store_name, provider_name, provider_package);
   

   Note: provider_package is the dbms_dbfs_hs package.

4. Mount the stores for access using:

   DBMS_DBFS_CONTENT.mountStore(store_name, store_mount, singleton,principal,
     owner, acl, asof, read_only);

Use the DBMS_DBFS_CONTENT package to create, update, read, and delete file system entries in the store.

You can drop a hierarchical store.

DBMS_DBFS_HS.dropStore(store_name, opt_flags);

The DBFS hierarchical store can store its files in compressed forms.

The DBFS hierarchical store has the ability to store its files in compressed form using the SETPROPERTY method and the property PROPNAME_COMPRESSLVL to specify the compression level.

Valid values are:

• PROPVAL_COMPLVL_NONE: No compression

• PROPVAL_COMPLVL_LOW: LOW compression

• PROPVAL_COMPLVL_MEDIUM: MEDIUM compression

• PROPVAL_COMPLVL_HIGH: HIGH compression

Generally, the compression level LOW performs best and still provides a good compression ratio. Compression levels MEDIUM and HIGH provide significantly better compression ratios, but compression times can be correspondingly longer. Oracle recommends using NONE or LOW when write performance is critical, such as when files in the DBFS HS store are updated frequently. If space is critical and the best possible compression ratio is desired, use MEDIUM or HIGH.

Files are compressed as they are paged out of the cache into the staging area (before they are subsequently pushed into the back end tape or S3 storage). Therefore, compression also benefits by storing smaller files in the staging area and effectively increasing the total available capacity of the staging area.

This example program configures and uses a tape store.

In the example, you must substitute valid values in some places, as indicated by <...>, for the program to run successfully.

Rem Example to configure and use a Tape store.
Rem
Rem hsuser should be a valid database user who has been granted
Rem the role dbfs_role.
 
connect hsuser/hsuser 
 
Rem The following block sets up a STORETYPE_TAPE store with
Rem DBMS_DBFS_HS acting as the store provider.
 
declare 
storename varchar2(32) ; 
tblname varchar2(30) ; 
tbsname varchar2(30) ; 
lob_cache_quota number := 0.8 ; 
cachesz number ; 
ots number ; 
begin 
cachesz := 50 * 1048576 ; 
ots := 1048576 ; 
storename := 'tapestore10' ; 
tblname := 'tapetbl10' ; 
tbsname := '<TBS_3>' ; -- Substitute a valid tablespace name
 
-- Create the store.
-- Here tbsname is the tablespace used for the store,
-- tblname is the table holding all the store entities,
-- cachesz is the space used by the store to cache content
--   in the tablespace,
-- lob_cache_quota is the fraction of cachesz allocated
--   to level-1 cache and
-- ots is minimum amount of content that is accumulated
--   in level-2 cache before being stored on tape
dbms_dbfs_hs.createStore(
  storename,
  dbms_dbfs_hs.STORETYPE_TAPE,
  tblname, tbsname, cachesz,
  lob_cache_quota, ots) ; 
 
dbms_dbfs_hs.setstoreproperty(
  storename,
  dbms_dbfs_hs.PROPNAME_SBTLIBRARY,
  '<ORACLE_HOME/work/libobkuniq.so>') ;
  -- Substitute your ORACLE_HOME path
 
dbms_dbfs_hs.setstoreproperty(
  storename,
  dbms_dbfs_hs.PROPNAME_MEDIAPOOL,
  '<0>') ;  -- Substitute valid value
 
dbms_dbfs_hs.setstoreproperty(
  storename,
  dbms_dbfs_hs.PROPNAME_COMPRESSLEVEL,
  'NONE') ; 
 
-- Please refer to DBMS_DBFS_CONTENT documentation
-- for details about this method
dbms_dbfs_content.registerstore(
  storename,
  'tapeprvder10',
  'dbms_dbfs_hs') ; 
 
-- Please refer to DBMS_DBFS_CONTENT documentation
-- for details about this method
dbms_dbfs_content.mountstore(storename, 'tapemnt10') ; 
end ; 
/ 
 
Rem The following code block does file operations
Rem using DBMS_DBFS_CONTENT on the store configured
Rem in the previous code block
 
connect hsuser/hsuser 
 
declare 
  path varchar2(256) ; 
  path_pre varchar2(256) ; 
  mount_point varchar2(32) ; 
  store_name varchar2(32) ; 
  prop1 dbms_dbfs_content_properties_t ; 
  prop2 dbms_dbfs_content_properties_t ; 
  mycontent blob := empty_blob() ; 
  buffer varchar2(1050) ; 
  rawbuf raw(1050) ; 
  outcontent blob := empty_blob() ; 
  itemtype integer ; 
  pflag integer ; 
  filecnt integer ; 
  iter integer ; 
  offset integer ; 
  rawlen integer ; 
begin 
 
  mount_point := '/tapemnt10' ; 
  store_name := 'tapestore10' ; 
  path_pre := mount_point ||'/file' ; 
 
 
-- We create 10 empty files in the following loop
  filecnt := 0 ; 
  loop 
    exit when filecnt = 10 ; 
    path := path_pre || to_char(filecnt) ; 
    mycontent := empty_blob() ; 
    prop1 := null ; 
 
    -- Please refer to DBMS_DBFS_CONTENT documentation
    -- for details about this method
    dbms_dbfs_content.createFile(
      path, prop1, mycontent) ; -- Create the file
 
    commit ; 
    filecnt := filecnt + 1 ; 
  end loop ; 
 
  -- We populate the newly created files with content
  -- in the following loop
  pflag := dbms_dbfs_content.prop_data +
           dbms_dbfs_content.prop_std  +
           dbms_dbfs_content.prop_opt  ; 
 
  buffer := 'Oracle provides an integrated management '  ||
            'solution for managing Oracle database with '||
            'a unique top-down application management '  ||
            'approach. With new self-managing '          ||
            'capabilities, Oracle eliminates time-'      ||
            'consuming, error-prone administrative '     ||
            'tasks, so database administrators can '     ||
            'focus on strategic business objectives '    ||
            'instead of performance and availability '   ||
            'fire drills. Oracle Management Packs for '  ||
            'Database provide signifiCant cost and time-'||
            'saving capabilities for managing Oracle '   ||
            'Databases. Independent studies demonstrate '||
            'that Oracle Database is 40 percent easier ' ||
            'to manage over DB2 and 38 percent over '    ||
            'SQL Server.'; 
 
  rawbuf := utl_raw.cast_to_raw(buffer) ; 
  rawlen := utl_raw.length(rawbuf) ; 
  offset := 1 ; 
  filecnt := 0 ; 
  loop 
    exit when filecnt = 10 ; 
    path := path_pre || to_char(filecnt) ; 
    prop1 := null;
  
    -- Append buffer to file
    -- Please refer to DBMS_DBFS_CONTENT documentation
    -- for details about this method
    dbms_dbfs_content.putpath(
      path, prop1, rawlen,
      offset, rawbuf) ;
 
    commit ; 
    filecnt := filecnt + 1 ; 
  end loop ; 
 
  -- Clear out level 1 cache
  dbms_dbfs_hs.flushCache(store_name) ; 
  commit ; 
 
  -- Do write operation on even-numbered files.
  -- Do read operation on odd-numbered files.
  filecnt := 0 ; 
  loop 
    exit when filecnt = 10; 
    path := path_pre || to_char(filecnt) ; 
    if mod(filecnt, 2) = 0 then 
      -- Get writable file
      -- Please refer to DBMS_DBFS_CONTENT documentation
      -- for details about this method
      dbms_dbfs_content.getPath(
        path, prop2, outcontent, itemtype,
        pflag, null, true) ;
  
      buffer := 'Agile businesses want to be able to '    ||
                'quickly adopt new technologies, whether '||
                'operating systems, servers, or '         ||
                'software, to help them stay ahead of '   ||
                'the competition. However, change often ' ||
                'introduces a period of instability into '||
                'mission-critical IT systems. Oracle '    ||
                'Real Application Testing-with Oracle '   ||
                'Database 11g Enterprise Edition-allows ' ||
                'businesses to quickly adopt new '        ||
                'technologies while eliminating the '     ||
                'risks associated with change. Oracle '   ||
                'Real Application Testing combines a '    ||
                'workload capture and replay feature '    ||
                'with an SQL performance analyzer to '    ||
                'help you test changes against real-life '||
                'workloads, and then helps you fine-tune '||
                'the changes before putting them into'    ||
                'production. Oracle Real Application '    ||
                'Testing supports older versions of '     ||
                'Oracle Database, so customers running '  ||
                'Oracle Database 9i and Oracle Database ' ||
                '10g can use it to accelerate their '     ||
                'database upgrades. '; 
 
      rawbuf := utl_raw.cast_to_raw(buffer) ; 
      rawlen := utl_raw.length(rawbuf) ; 
 
      -- Modify file content
      -- Please refer to DBMS_DBFS_CONTENT documentation
      -- for details about this method
      dbms_lob.write(outcontent, rawlen, 10, rawbuf);
      commit ; 
    else 
      -- Read the file
      -- Please refer to DBMS_DBFS_CONTENT documentation
      -- for details about this method
      dbms_dbfs_content.getPath(
        path, prop2, outcontent, itemtype, pflag) ;
    end if ; 
    filecnt := filecnt + 1 ; 
  end loop ; 
 
  -- Delete the first 2 files
  filecnt := 0; 
 
  loop 
    exit when filecnt = 2 ; 
    path := path_pre || to_char(filecnt) ; 
    -- Delete file
    -- Please refer to DBMS_DBFS_CONTENT documentation
    -- for details about this method
    dbms_dbfs_content.deleteFile(path) ;
    commit ; 
    filecnt := filecnt + 1 ; 
  end loop ; 
 
 
  -- Move content staged in database to the tape store
  dbms_dbfs_hs.storePush(store_name) ; 
  commit ; 
 
end ; 
/

DBFS Links allows storing SecureFiles LOBs transparently in a location separate from the segment where the LOB is normally stored. Instead, you store a link to the LOB in the segment.

The link in the segment must reference a path that uses DBFS to locate the LOB when accessed. This means that the LOB could be stored on another file system, on a tape system, in the cloud, or any other location that can be accessed using DBFS.

When a user or application tries to access a SecureFiles LOB that has been stored outside the segment using a DBFS Link, the behavior can vary depending on the attempted operation and the characteristics of the DBFS store that holds the LOB:

• Read:

  If the LOB is not already cached in a local area in the database, then it can be read directly from the DBFS content store that holds it, if the content store allows streaming access based on the setting of the PROPNAME_STREAMABLE parameter. If the content store does not allow streaming access, then the entire LOB will first be read into a local area in the database, where it will be stored for a period of time for future access.

• Write:

  If the LOB is not already cached in a local area in the database, then it will first be read into the database, modified as needed, and then written back to the DBFS content store defined in the DBFS Link for the LOB in question.

• Delete:

  When a SecureFiles LOB that is stored through a DBFS Link is deleted, the DBFS Link is deleted from the table, but the LOB itself is NOT deleted from the DBFS content store. Or it is more complex, based on the characteristics/settings, of the DBFS content store in question.

DBFS Links enable the use of SecureFiles LOBs to implement Hierarchical Storage Management (HSM) in conjunction with the DBFS Hierarchical Store (DBFS HS). HSM is a process by which the database moves rarely used or unused data from faster, more expensive, and smaller storage to slower, cheaper, and higher capacity storage.

Figure 22-1 Database File System Link

Description of "Figure 22-1 Database File System Link"

Database File System Links require the creation of a Database File System through the use of the DBFS Content package, DBMS_DBFS_CONTENT.

Oracle provides several methods for creating a DBFS Link:

• Move SecureFiles LOB data into a specified DBFS pathname and store the reference to the new location in the LOB.

  Call DBMS_LOB.MOVE_TO_DBFS_LINK()with LOB and DBFS path name arguments, and the system creates the specified DBFS HSM Store if it does not exist, copies data from the SecureFiles LOB into the specified DBFS HSM Store, removes data from the SecureFiles LOB, and stores the file path name for subsequent access through this LOB.

• Copy or create a reference to an existing file.

  Call DBMS_LOB.COPY_DBFS_LINK() to copy a link from an existing DBFS Link. If there is any data in the destination SecureFiles LOB, the system removes this data and stores a copy of the reference to the link in the destination SecureFiles LOB.

• Call DBMS_LOB.SET_DBFS_LINK(), which assumes that the data for the link is stored in the specified DBFS path name.

  The system removes data in the specified SecureFiles LOB and stores the link to the DBFS path name.

Creating a DBFS Link impacts which operations may be performed and how. Any DBMS_LOB operations that modify the contents of a LOB will throw an exception if the underlying LOB has been moved into a DBFS Link. The application must explicitly replace the DBFS Link with a LOB by calling DBMS_LOB.COPY_FROM_LINK() before making these calls.

When it is completed, the application can move the updated LOB back to DBFS using DBMS_LOB.MOVE_TO_DBFS_LINK(), if needed. Other DBMS_LOB operations that existed before Oracle Database 11g Release 2 work transparently if the DBFS Link is in a file system that supports streaming. Note that these operations fail if streaming is either not supported or disabled.

If the DBFS Link file is modified through DBFS interfaces directly, the change is reflected in subsequent reads of the SecureFiles LOB. If the file is deleted through DBFS interfaces, then an exception occurs on subsequent reads.

For the database, it is also possible that a DBA may not want to store all of the data stored in a SecureFiles LOB HSM during export and import. Oracle has the ability to export and import only the Database File System Links. The links are fully qualified identifiers that provide access to the stored data, when entered into a SecureFiles LOB or registered on a SecureFiles LOB in a different database. This ability to export and import a link is similar to the common file system functionality of symbolic links.

The newly imported link is only available as long as the source, the stored data, is available, or until the first retrieval occurs on the imported system. The application is responsible for stored data retention. If the application system removes data from the store that still has a reference to it, the database throws an exception when the referencing SecureFiles LOB(s) attempt to access the data. Oracle also supports continuing to keep the data in the database after migration out to a DBFS store as a cached copy. It is up to the application to purge these copies in compliance with its retention policies.

The API DBMS_LOB.COPY_DBFS_LINK(DSTLOB, SRCLOB, FLAGS) provides the ability to copy a linked SecureFiles LOB.

sBy default, the LOB is not obtained from the DBFS HSM Store during this operation; this is a copy-by-reference operation that exports the DBFS path name (at source side) and imports it (at destination side). The flags argument can dictate that the destination has a local copy in the database and references the LOB data in the DBFS HSM Store.

You can copy DBFS links from source tables to destination tables.

Online redefinition copies any DBFS Links that are stored in any SecureFiles LOBs in the table being redefined.

Online redefinition copies any DBFS Links that are stored in any SecureFiles LOBs in the table being redefined.

DBFS Links can read from a linked SecureFiles LOB even if the data is not cached in the database.

You can read data from the content store where the data is currently stored and stream that data back to the user application as if it were being read from the SecureFiles LOB segment. This allows seamless access to the DBFS Linked data without the prerequisite first call to DBMS_LOB.COPY_FROM_DBFS_LINK().

Whether or not transparent read is available for a particular SecureFiles LOB is determined by the DBFS_CONTENT store where the data resides. This feature is always enabled for DBFS_SFS stores, and by default for DBFS_HS stores. To disable transparent read for DBFS_HS store, set the PROPNAME_STREAMABLE parameter to FALSE.

The DBMS_DBFS_HS PL/SQL package constants are very detailed.

There are many methods in the DBMS_DBFS_HSpackage.

There are several views available for DBFS Hierarchical Store.

There are several views available for the DBFS Hierarchical Store.