Apr 6 ’11
The Mainframe DBA: Empowering Technicians for the Changing World
In a mainframe environment, many different types of Database Administrators (DBAs) are needed to support building and managing applications critical to a business. In smaller organizations, a DBA typically must wear many hats:
- As a database designer, he must learn design techniques and transform a logical data model that represents the business needs into a physical database that supports application processing needs.
- As an application DBA, he works with application developers to build efficient SQL for optimal performance.
- The DBA needs to know how to design and test database backup and recovery procedures for both on-site and off-site disaster recovery. DB2 Database Management System (DBMS) security is also usually handled by the DBA who must understand the security requirements for both the data and the applications accessing the data.
- In an ongoing support role, the DBA may need to troubleshoot both system and application performance issues as they arise. In some cases, these issues will require fixes to be applied. The DBA needs to access IBM support to open issues and get fixes to problems.
- The DBA must understand the differences between the multiple releases of DB2 and how they impact development and management procedures. He may even have to install and configure DB2 releases.
A DBA in a small organization has many varied responsibilities. In larger organizations, each of these specialty areas is typically split among different people, teams and, sometimes, different departments. It can take years to become an expert in any one of these jobs:
- The systems DBA handles installing and fixing the DBMS, along with configuring system resources to tune DBMS performance.
- The data modeler DBA will collect business requirements and design a logical database that supports the business.
- The application DBA will take the logical model and build the physical data structures the application needs. He’ll work with development to understand the application and SQL access, then design indexes to support these applications. Because he understands the applications using the database, the application DBA will usually design backup and recovery procedures needed in the production environment.
- The performance analyst is a DBA who has years of experience tuning the applications and database for the best balance between performance, response time, and system resource consumption. He works with developers to improve the overall performance of the application.
- The data warehouse DBA specializes in designing databases that support business intelligence reporting.
- If more than one variety of DBMS is installed, there are often people who specialize in DB2, Oracle, or other databases. You may also find specialty DBAs who work with Enterprise Resource Planning (ERP) applications such as PeopleSoft, Oracle, and SAP.
In a small organization, a DBA may learn how to perform a portion of the duties of each of these roles. However, it would be difficult to become an expert within any one role due to the time and effort it takes to perform the job. In larger organizations, the DBA will be assigned to focus on one area of responsibility and will become a specialist in this area.
The issue with this granularity of responsibilities is that the DBA doesn’t learn the other specialties. Because the DBA hasn’t worked with or learned the different specialties, it makes it difficult for him to be portable and help out in other areas where he has little experience. For example, an SQL performance and tuning DBA can’t help with a recovery of DB2 objects because he hasn’t learned how to find recovery points for the data modified by an application or how to run the recovery utility once the recovery scenario is defined. It would be difficult for the SQL tuning DBA because the recovery tools have different interactions, commands, and navigations from the SQL tuning tools.
Given the consistent capacity growth since 2000, the resurgence of System z sales since IBM released zEnterprise, and how well-suited the mainframe is for cloud computing, the mainframe will only become more critical and further embedded for many years to come.
As the number of transactions and quantity of data on the mainframe continues to multiply, it’s increasingly difficult to find skilled mainframe DB2 DBAs who know how to perform the many different tasks required to work with z/OS and the many different database management tools. Additionally, the current generation of computer science students has grown up with rich, graphical, browser-based applications with built-in collaboration capabilities, such as Facebook. This new generation of mainframe administrators will expect and demand these kinds of tools for the mainframe.
The challenge companies face today is finding modernized tools with rich visual graphics and collaboration to manage the mainframe. Companies will find it difficult to attract and retain the best and brightest college graduates with the existing tools and interaction model. The 3270 interface was never easy to learn; this accounted for the long learning curve of legacy mainframe professionals who started working with it in the ’70s and ’80s and became more accomplished over the years. Even if young mainframers had an interest in learning the 3270 interface, it could take them years to become proficient because of the many commands and functions spread across multiple tools. Companies need these people to be productive immediately.
DBAs at small companies must be the jack of all trades; they’re tasked with managing more mission-critical databases with fewer and fewer people. Where a separate toolset exists, each has its own interface, commands, and ways of interpreting the data. The job has to get simpler, which means the tools and technology the DBA uses must be improved. Business data is simply too critical to neglect any of the many DBA roles and responsibilities.
How can you make this job easier—not just for the overburdened experts but also for the new kids on the block? The right answer would be to simplify and streamline existing mainframe management tasks, increase staff productivity, and facilitate collaboration and knowledge sharing to reduce the cost and complexity of managing strategic mainframe workloads well into the 21st century. This solution would also need a way to integrate existing tools because it isn’t likely that any company could completely begin anew. Sounds almost impossible, doesn’t it?
A portal could be an answer, but it wouldn’t go far enough because a portal is really just skin deep. An entirely new workspace is needed to shorten the on-ramp for the next generation of mainframe workforce while also offering significant productivity improvements to today’s mainframe experts. Existing capabilities of the various tools would have to be integrated into the workspace, but the functionality for each DBA role would have the same look and feel, facilitating a fluid migration between roles. Ideally, particularly for larger shops, the workspace would need to facilitate collaboration and knowledge sharing to accelerate mainframe workforce productivity. Too many mainframers work in isolation; a mainframer who leverages teamwork and collaboration can reap immediate, tangible results in terms of productivity, fulfilling Service Level Agreements (SLAs) and controlling costs.
What would this new workspace have in terms of features? If you could start over, perhaps you would look for:
- Customized workspace
- Process capture “wizards”
- Collaboration with team and other experts internally and externally
- In-context knowledge from experts
- Rich visual presentation
- Historical data at hand
- Graphing and comparative history
- Automated problem solving.
The management solution would capture knowledge from the experienced worker to be used by the new administrator while also helping the experienced administrator. The workspace would provide annotation capability on captured processes for team sharing. The workspace would be customizable with the ability to place tools on an individualized workspace to perform the task in the way the user wishes to work. Users would be able to access real-time data and review the historical information; they could graph and compare this information for different times. Additionally, while working on a given problem, they could access in-context help from other experts in the field.
With such a solution, a company would reap huge benefits in the efficiency and performance of its DB2 systems and its employees. Rich visual presentation of management data would make arcane knowledge more readily available to a new generation of workers and would include representations of relationships—which databases are used by which processors and Logical Partitions (LPARs). This would lead to faster resolution times and more uptime and faster on-ramping of less experienced DBAs.
Third-party integration of domain documentation and expertise for in-context guidance would decrease persistent context switching from the mainframe to the Internet, to email, to instant messaging, etc. This would improve efficiency. Intelligent alerting designed to focus attention on real issues and automation wherever possible to minimize the rote work required would free up time for more complex endeavors. These are just some of the benefits to be realized from a new intelligent workspace for DB2 database management.
Companies can’t afford to hire larger staffs, even as the workload grows, the cost of downtime soars, and mainframe experts begin to retire. Software vendors need to work with their customers in new and different ways to address these needs. They need to do so by not simply updating their solutions, but by offering something new and better. Companies need to take ownership of the solution, too, and embrace new technology. It will take time and the right partners, but the first step is to base the new solution on a suite of intelligent existing tools that support the DB2 DBA. Working together, vendors and customers can achieve the comprehensive solution outlined here and ensure their business—supported by mainframes—is managed efficiently and effectively well into the future.