IBM Confidential: System/360 File Organization [video]

### TL;DR IBM System/360 file organization covers sequential and random data, supports all processing modes via indexing, and provides device independence and cost advantages. ## Relationship Between File Organization and Data Processing A file can be defined as "file the file in the file" — the first "file" is the action, the second is the content, and the third is the device. This article focuses on the content level. Any application falls into one of four categories: - Sequential data + sequential processing - Sequential data + random processing - Random data + sequential processing - Random data + random processing ## Processing Methods Under Sequential Data Organization ### Sequential Processing Transaction data traverses the file in order (e.g., payroll processing: from employee #1 to the last, generating checks one by one). ### Skip-Sequential Processing (Batch) Still moves forward sequentially but skips intermediate records. For example, process records 4, 6, 9 first, then jump to 13. Batches can be large or small. Using tape for small batches requires traversing the entire file; using direct-access storage devices requires only hits × transaction time. ### Random Processing Jumps back and forth within a sequential file: from record 6 back to 2, then to 10, 14, back to 7. This supports IBM remote processing devices — data can be processed instantly via remote input, with real-time query support. ## Processing Methods Under Random Data Organization Even if the data file is random internally, sequential processing is still possible (starting from record 1, then 2, 3, 4...). Random organization with random processing is a mixed scenario. IBM data storage devices cover all categories. ## Indexing: The Core Tool for Unified Organization An index is a sorted reference list of file contents, similar to a book or road map. For example, finding Flint, Michigan: first check the index to locate the state in region K4, then check the region index to find the city in region D5 — this is a two-level index example. ### Indexing Applied to Data Files Data files span multiple tracks. Example index: - Record 5: index shows 5 > 1 and < 106, so it's on track 2. - Record 109: index shows > 106 and < 211, so it's on track 3. ### Advantages of Indexing - Enables uniform organization across all direct-access storage devices (2311, 2302, 2321): a single program works with different devices, easing installation, programming, and device conversion. - Supports all processing activities: sequential, skip-sequential, random processing, random queries, remote processing — all achievable on the same data file via indexing. ## Device Independence and Programming Support IBM provides comprehensive programming support to ensure device independence: a program written for one storage device can be used with another. Supports COBOL, FILE statements in multipurpose programming languages, System/360 sort utilities, and full data management support. This advantage is unmatched by competitors. ## Historical Cost Reduction Experience IBM has deep experience in data storage devices, especially disks. Starting with the 305 system in 1956, cost per million characters varied with capacity. The 1405 then lowered costs, and successive generations of devices (up to the 2321 data cell in 1964) continuously reduced cost per million characters while improving performance. Salespeople and customers should understand this trend. ## Unique Advantages of System/360 ### Load-and-Format Formatting is automatic when writing new records; no programmer pre-planning of record layout is needed. The system handles format automatically after input. ### Keyed Records and Remote Processing When receiving multiple types of information via remote processing devices, previously separate tape drives or prior sorting were required. Now, information can be written directly to disk, with segmentation enabling fast access. For output, when the system is ready and a printer is idle, information can be retrieved from a specific disk area to print reports. ### Symbolic Allocation of Storage Space After writing information on a disk pack or data cell, it can be removed for storage. When processing again, there is no need to return to the original drive; it can be mounted on any other drive or even moved to another system. In summary, IBM offers a full line of products covering different data rates and access times, capable of handling large and small records, with corresponding programming support. The next steps depend on understanding the customer's records and data processing needs. Good luck with sales! --- **Source:** [IBM Confidential: System/360 File Organization [video]](https://www.youtube.com/watch?v=zokKqP0plrM)