IBM recently published reference architecture blueprints to simplify Tivoli Storage Manager (TSM) sizing and deployment, “Blueprint and Server Automated Configuration for Linux x86". Feedback from IBM Business Partners has been positive. In December 2013, Version 1.2 was released for free download.
What is a Reference Architecture Blueprint?
A reference architecture blueprint is a detailed hardware specification designed to manage a specific workload. In this case, the blueprints specify hardware requirements for small, medium and large TSM workloads, using Best Practices for data protection.
Why Use Reference Architecture Blueprints Instead of Sizing?
Traditional backup server sizing is extremely flexible and recommended for unique workloads or where precision is critical. Most of the time, however, a reference architecture blueprint can reduce time and complexity, and improve data protection through standardization and automation.
IBM reference architecture blueprints for TSM have companion configuration scripts that set up TSM for the requested workload quickly and consistently.
I supported TSM for 5 years as an IT Specialist Manager, and have seen many clients struggle with traditional backup server sizing. Even if you can capture details about the current backup workload, which is often impossible, you have to estimate the data growth rate. Sizing errors can result in unhappy application owners or an inability to meet recovery objectives. In contrast, a reference architecture blueprint takes you to a known destination quickly and safely.
Recently, TSM experts had a TSM sizing discussion on ADSM.org. Baltimore-Washington TSM User Group Member Wanda Prather, Stefan Folkerts, and Sergio Fuentes discussed sizing variables: Workload, hardware, storage features. As Wanda says, sizing is, “both an ‘it depends’ and a ‘what not to do’ answer.” I recommend reading her post if you’re interested in backup server sizing.
Reference Architecture Blueprint Details
These reference architectures are based on x86 hardware running Linux and are optimized as disk only storage using TSM data deduplication technology. They have been tested to determine the optimal workloads and limits for each size: Small, Medium or Large. The blueprint includes a document, or “Cookbook”, that describes the three reference architectures in detail, including IBM hardware model numbers and configuration requirements.
IBM also includes two scripts to speed up the installation and configuration, increasing time-to-value:
The first script does a configuration check that will verify the hardware configuration meets the blueprint specifications, validate Linux kernel settings, and verify the configuration of required file systems prior to running the standard TSM Server installation. The script also configures the TSM Server using best practices and: creates a DB2® instance, defines deduplication storage pools with optimal performance settings, defines administrative maintenance tasks optimized for de-duplication scalability, defines TSM database backup to disk, creates a dsmserv.opt file with best practice option overrides, creates policy domains for database, mail and file servers with management classes for 30-, 60- and 120-day retention, and defines backup schedules for all client types that can be easily selected when deploying the desired client workloads.
The second script runs simulated TSM database and storage pool workloads and provides performance measurements that can be used to compare as a reference against those measured on the blueprint configuration.
Please share your experiences with TSM reference architecture blueprints and backup server sizing. What works for you?
Mike Barton is a Worldwide Storage Marketing Manager for IBM.
The opinions expressed herein are sorely mine.