Browse Prior Art Database

Method to Avoid Replication Delays for Large Update Queries

IP.com Disclosure Number: IPCOM000245749D
Publication Date: 2016-Apr-04
Document File: 1 page(s) / 26K

Publishing Venue

The IP.com Prior Art Database

Abstract

Described is a method to avoid replication delays for large update queries.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 1

Method to Avoid Replication Delays for Large Update Queries

Many modern data computer systems that run production applications involve data replication. Data replication is, as the name implies, a methodology to replicate data from one system to another. This work is usually done for disaster recovery purposes but is also done to create a replica database that can be queried.

    With SQL it can be easy to write a simple statement that affects a large amount of data. Recently, an IBM* i customer experienced a problem where a query was run that updated tens of millions of rows in a table. The query, itself, ran in a short amount of time, but the customer's data replication fell behind by over 12 hours due to having to replicate so many changes from the source machine to the target machine. In this case, the target machine was their disaster recovery server located several thousand miles away. The combination of the amount of changes and longer network time left them at risk of data loss for many hours until they caught up. Often times, as in this case, the end user or programmer does not realize the query would affect their environment in such a drastic way.

    As to the second reason for data replication, savvy customers have long ago realized that allowing ad hoc queries and generally querying a large production database that is being constantly updated can lead to undesired consequences. Therefore, a backup database is created such that end users can query a replica database, thereby offloading the additional processing to separate machine(s).

    There already exists query governor technology to stop queries from running if they are expe...