File architecture overview

From m204wiki
Revision as of 13:46, 23 May 2013 by Lrenton (talk | contribs)
Jump to navigation Jump to search

Summary

Model 204 files provide a highly flexible environment for handling large or small amounts of data. The system supports all types of data structures: flat structures, relations, hierarchies, and networks.


Model 204 uses inverted file retrieval techniques. These techniques facilitate rapid retrieval of data without requiring expensive scanning of the database itself.

Data within a Model 204 file are kept in an arbitrary collection of Records.

Records in a file normally are related to each other, although they need not be. Records within a given file may have the same format (same collection of field names) or any number of different formats (containing any mixture of the fields defined to Table A, below). A file can contain as many as 16.7 million records. With the new FILEORG x’200’ setting, available from Model 204 V7R5 onwards, this upper limit will increase to 48 million.

Files can be linked together logically according to field values within the files. Any number of files can be linked in this way. As many as 32,767 files can be accessed in one Model 204 job.

Model 204 files consist of one or more data sets. Each data set is formatted into fixed-length physical records called pages, and all pages are the same size: 6184 bytes.

Internally, these pages are organized into tables: the File Control Table (FCT); Table A; Table B; Table C; Table D; Table E; and Table X. Any pages in the file not allocated to one of these tables are considered free space. See Managing File and Table Sizes for how this free space can be applied to the other tables.

The components of a Model 204 File



File Control Table

The File Control Table (FCT) Contains file parameter settings, data set or file definition names of all data sets in the file, the status of the file, and other control information. The FCT is always 8 pages.


The Internal File Directory

Table A contains three structures:

A dictionary of the fieldgroup / field names and their attributes.

Some attributes (notably 'CODED') require lists of values to be maintained. These lists are stored either in the FEW-VALUED or MANY-VALUED structures.

Table A usually is small in relation to the rest of the file. The field name section in particular should be as small as possible to aid efficient access, especially if your site uses field name variables


Data

Data Structures

All data held in a Model204 file are held in Records.

Model 204 records are best thought of as very loosely defined containers, with almost no fixed structure


Record Collection of fields (either individually or in Repeating Field Groups (RFGs).

Each record is variable in length and need contain only the fields that pertain to it. The limit of the number of field value pairs in a record is in the tens of millions.

There is only a limited fixed format for a record (preallocated fields). Almost any number of fields can appear almost any number of times in almost any order. Each record is automatically assigned a unique internal record number that is used by the system to build index entries for the record.

Repeating Field Group (RFG) Available as of Model 204 V7R5.

A Repeating Field Group (RFG) is, as the name implies, a set of fields which are retrieved and updated in a single operation.

Field Elementary data item.

Fields are defined with attributes that control storage formats and indexing options. Up to 4000 (32000 as of Model 204 V7R5, when using FILEORG x'100') different field names can be defined in a single Model 204 file.

Data Tables

Table B contains (at least) the base data of all the records in the file. These base records contain pointers to any extensions which exist (whether they are elsewhere in Table B or in Table X).

Table X when enabled, holds all extension records. Table B is then used to store only base records, thus, maximizing the possible number of records can be stored regardless of the number of extension records.

Table E (when enabled) contains Large Object data for the file. The pointer to the starting point of the LOB is contained in Table B or X.


Indexing

Index Structures

The indexing structures necessary for direct retrieval of records are contained in Tables C and D.

There are two types of indexes, both of which utilize Table D List and Bitmap Pages:

Index Type Description
Ordered Index Is composed of the Ordered Index B-tree, contained in Table D, for ORDERED fields along with either a number of IMMEDiate pointers to the record in Table B,
or to a secondary index (of list or bitmap pages) located in Table D.
Hashed Index Is composed of Table C, which indexes KEY and NUMERIC RANGE values in fields, along with either a single
direct pointer (per segment) to a base record in Table B or or a secondary index (of list or bitmap pages) located in Table D.

Index Tables

Table C contains a series of entries for every field / value combination that occurs in the file for fields defined as KEY. There are also a series of entries for every Field Value Pair that occurs for fields that have the NUMERIC RANGE attribute.

The series consists of an entry cell, chained to cells with an entry for any segment that contains one or more occurrences of the Field Value Pair. If a Field Value Pair is unique within a segment, the Table C cell contains the IRN of the Table B record in which the field occurs. If the field name = value pair is not unique in the segment, the Table C cell contains a pointer to a bitmap or list page in Table D (as described above).


Table D contains: the Ordered Index B-tree node pages, which contain the values and record accessing information for all ORDERED fields; as well as the bitmap and list pages described above.

This includes the Existence Bit Map

Table D can be expanded with the INCREASE command.

Procedures and Miscellaneous Structures

Table D also contains a few other structures:

It contains procedure (code)and structures used to mange them: a procedure dictionary (used to store procedure names and aliases); and an Access Control Table (used to map user and procedure classes).

It contains a reserved area (DPGSRES) that primarily provides additional space for the transaction back out facility.

A map of the Preallocated Fields (used whenever a new records is stored) is also contained in Table D. This map can be seen by the DISPLAY RECORD command.