Surety is performing system maintenance this weekend. Electronic date stamps on new Prior Art Database disclosures may be delayed.
Browse Prior Art Database

Data aware optimized backups – Object & System level

IP.com Disclosure Number: IPCOM000246384D
Publication Date: 2016-Jun-02
Document File: 6 page(s) / 79K

Publishing Venue

The IP.com Prior Art Database


Mainframe DB2 (DB2 for z/OS) databases house thousands of objects that are accessed by mainframe and distributed applications. The size of such big data stores measure up to several Terabytes. The following are the most common backup / recovery techniques employed in DB2 for z/OS. • Traditional (Image) copy method using DB2 utilities. These are time consuming and recoveries can be slow. • SNAP shot method to take system level backups using technologies like IBM flash copy. They offer a high speed backup and recovery solution but the storage space requirements can go up drastically based on stringent data recovery requirements. The proposed method “Data aware Smart Backup” bridges the gap between the slow and steady utility based image copy process and the high speed flash copy process. It aims to achieve high speed database backup processing at optimal storage costs. Flash copy based backups also help in faster database recoveries thereby contributing towards high availability.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

Page 01 of 6

Data aware optimized backups - Object & System level

Salient features:

The proposed method has the following salient features:

1. A novel backup approach to look at changes on object(s) at a RDBMS level unlike existing methods that track changes at a disk hardware level. This is achieved by recording snapshot copies taken at an object/system level and using the RDBMS services to collect object update statistics in a meta-data store since the last snapshot copy. The meta-data store will be queried by the smart backup agent based on a filter criteria to identify candidate objects for backup.

2. A smart backup agent that will be able to identify the need for system level backups and incremental backups based on database analytics that will result in reduced backup and recovery times.

3. A process whose backup filter criteria can be customized to specially cater to critical objects.

Design :

The design of the proposed method is described in the below section:

A data aware backup solution that makes use of snapshot copy technology to backup objects at a system level or individual object level based on analytics. The existing metadata infrastructure in DB2 for z/OS will be leveraged to collect additional analytics information needed to make snapshot copies at an object/system level data aware.

    1. Base system level backup (Base SLB) - A one-time system level backup of the entire system that creates a reference point for performing analytics.

2. Augment meta-data infrastructure - Add additional columns and table(s) to Real time statistics (RTS) for DB2 to collect growth statistics on objects with respect to snapshot copies.

3. Data awareness - Only objects that require a backup to justify faster recovery times will be considered for individual snapshot copies.

4. Scalability - Solution scales well to bring in more optimization in the form of identifying critical objects for backup.

Acronyms that will be used in the following sections are as follows:

RDBMS - Relational Database Management System RTS - Real Time Statistics
SLB - System Level Backup
RBA - Relative Byte Address
PIT- Point In Time
ICTYPE - Type of Image copy (Backup)

Needed configuration:


Page 02 of 6

1. The Meta data table SYSIBM.SYSCOPY records the execution of utilities on objects like Image copy, load, REORGs etc. The proposed method will create an entry in this table whenever a SLB is taken.

* * * O 1234AAAAAA 2016-01-30-01.10.10

ICTYPE of 'O' indicates a SLB and the START_RBA indicates the starting point for all updates since the SLB was taken.

2. The real time statistics table SYSIBM.SYSTABLESPACESTATS collects various statistics on objects dynamically. The proposed method will add the following new columns to this table. These columns are to be maintained by RTS for every object in the system.

SLBLASTTIME : Timestamp of the last SLB

SLBUPDATEDPAGES : The number of distinct pages/blocks that...