Browse Prior Art Database

Improved VSAM Performance and Reliability Through Replacable Dummy High Key

IP.com Disclosure Number: IPCOM000196380D
Publication Date: 2010-Jun-02
Document File: 1 page(s) / 22K

Publishing Venue

The IP.com Prior Art Database

Abstract

Today with keyed VSAM data sets, the high key value must be maintained in several places including internal control blocks and the index component. When this data is shared between multiple users, data set breakage can result when this high key value is not updated properly do to timing errors or invalid serialization. Constantly having to update this field can also cause additional overhead. Users sometimes manually compute a high key and then add it to the data set as a dummy record, but this then prevents them from using this key as a legitimate key within the data set. Our new methodology will allow users to have an automatic dummy key created and added at the end of the data set. We will also keep track of the fact that this value is a dummy key, so that if the user ever adds a real record with this key value, it will be replaced without giving a duplicate key error.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 61% of the total text.

Page 1 of 1

Improved VSAM Performance and Reliability Through Replacable Dummy High Key

When you add records in that have an ascending key value to a VSAM data set, you are always updating the high key. Since the high key value is stored in the index, and internal control blocks, we always have to update the index and control blocks with the new high key value. If are adding a key that is lower than the high key, and can fit into an existing Control Interval (CI, essentially a VSAM block), then the index record does not have to be updated. The index points to the highest key within a CI block. So, if you are adding a record with a lower key to that block and it has room for the record, then the index does not have to be updated. Eventually when the CI block fills up, a new CI will be created during normal split processing. The high key value will then be moved to the new CI with available freespace.

At initial load, either VSAM or an OEM performance product will calculate the maximum key value for the data set, and insert as a record with that key value. A new bit will also be set on in the Catalog/VVDS to indicate that this data set contains a dummy record. Later if the user ever tries to add a record with that same key value, if the dummy record bit is on, that dummy record will be replaced with the users record, and the dummy record bit will be turned off. Any read activity that would read all of the records in the data set will ignore the highest key value record if the...