User control of packet generation

When backing up BS2000 files, the user can initiate a further subdivision of the automatically generated packets using the SELECT=*FROM-FILE(...) option.

Automatic packet division by HSMS (see section "Automatic packet generation") also permits parallel operation in cases where in earlier HSMS versions the user – as described below – had to control packet generation. In addition, automatic packet division by HSMS is more favorable for parallel operation than user-controlled packet generation because the location of the files on autonomous disks which is important for parallel operation is taken into account internally.

Whenever a user has many large files in a system, it can be advantageous to use several parallel subtasks to save these files. How this is done is described below.

When saving, it is specified in the input or except file which user files are to be saved or excluded from the save and in which sequence they should be processed.

Files with different user IDs or different catalog IDs are divided into different packages. In addition, files with the same user and catalog identification are split up into a maximum of 4 packages depending on their disk position. These packages are distributed to parallel subtasks and processed in parallel. The mechanisms provided by HSMS or ARCHIVE are usually completely sufficient for an optimal operation.

In rare exceptional cases, e.g. when backing up a few but very large files, however, it may make sense to use the following auxiliary construction in order to achieve an even utilization of the drives:

Every record in an input or except file must contain a path name. It may also optionally contain a group ID. The group ID is used to distribute files to separate packet for parallel processing by different subtasks. Specifying the group ID does not, however, allow files from different user or catalog IDs to be forced into the same packet!

[<gruppen-id>;] <pfadname>

The path name can be specified as a fully or partially qualified name with or without a catalog ID and a user ID. The use of wildcards is allowed. If a group ID is specified, it must be at least 1 but no more than 8 alphanumeric characters in length. If no group ID is specified, a blank is assumed.

If group IDs are used together with file excepts (EXCEPT-FILE-NAMES operand), the group IDs specified in the except file must be compatible with the group IDs specified in the selection file. 

Examples

The following examples illustrate the conversion of an HSMS file specification to an ARCHIVE specification (NAME operand).

Example 1

Input file:	:a:file1 :a:file2 :a:file3 :b:file1
ARCHIVE entries:	NAME 1 :a:file1 :a:file2 :a:file3	NAME 2 :b:file1

Example 2

Input file:	pkt1;:a:file1 pkt2;:a:file2 pkt1;:a:file3 pkt1;:b:file1
ARCHIVE entries:	NAME 1 :a:file1 :a:file3	NAME 2 :a:file2	NAME 3 :b:file1

Example 3

Input file:	:a:file1 2;:a:file2 :a:file3 1;:b:file1
ARCHIVE entries:	NAME 1 :a:file1 :a:file3	NAME 2 :a:file2	NAME 3 :b:file1

Example 4

Input file:	PKT1;:a*:$u*.file1 PKT2;:a*:$u*.file2	Pubset on the system: a1, a2
ARCHIVE entries:	NAME 1 :a1:$u*.file1	NAME 2 :a1:$u*.file2
	NAME 3 :a2:$u*.file1	NAME 4 :a2:$u*.file2

Example 5

Input file:	pkt1;:a:file1* pkt2;:a:file2* pkt1;:b:file1*
Except file:	pkt1;:a:file13 pkt2;:a:file22 pkt1;:b:file11
ARCHIVE entries:	NAME1 :a:file1* except :a:file13
	NAME2 :a:file2* except :a:file22
	NAME 3 :b:file1* except :b:file11