Browse Prior Art Database

Detection of Inconsistent Control Interval (torn page)

IP.com Disclosure Number: IPCOM000247899D
Publication Date: 2016-Oct-10
Document File: 2 page(s) / 39K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method to manage the exposure to an inconsistent Control Interval (CI), within a 64-bit operating system for mainframes, in an application and middleware- independent manner.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 2

Detection of Inconsistent Control Interval (torn page)

In a 64-bit operating system for mainframes, the unit of transfer for input/output (IO) for a Virtual Storage Access Method (VSAM) data set is known as a Control Interval (CI). Generally, a CI is represented on disk as a single record. However, in order to make efficient usage of disk space, certain CI sizes may result in more than one block (or record) on disk.

Figure: For a 3390 disk, the following CI sizes will result in more than one block per CI

When writing updates to a CI, the associated operating system always writes whole CIs in a single channel program; no channel program is built to only write an update to part of a CI. Due to some unlikely but possible error conditions (e.g., power loss), a CI consisting of multiple blocks (or records) may become inconsistent. This can occur when the first block(s) of a CI contain new updated data, but the last block(s) of the CI have the old non-updated data. This condition is known as an inconsistent CI or a torn page.

Currently, no explicit method exists for detecting this inconsistent CI unless the application has embedded meta-data in the user data portion of the CI to detect inconsistencies. Reading undetected inconsistent CIs is a data integrity issue (data integrity has been lost).

The novel solution is a method to manage the exposure to an inconsistent CI in an application and middleware- independent manner.

1


Page 02 of 2

The construction of the token based on a portion of the Time-of-Day (TOD) clock is unique. The solution uses the whole TOD clock value, which provides additional benefits such as an even lower probability of missing an error and a

mechanism for determining when the records were actually written (i.e., when the damage occurred). The default behavior is to perform the CI consistency checks. This solution imbeds the de...