For over 50 years the IBM mainframe has played a pivotal role in large-scale business data processing. Have you used an ATM or credit card today, or ordered a product online? You get the idea: Mainframes are everywhere and continue to be the platform of choice for massive processing efforts.
The IBM DB2 Analytics Accelerator for z/OS
is a workload optimized appliance for the z Systems platform released in 2012. It was designed to create faster business insights by utilizing stored data to quickly provide answers. Truly, it's a technical solution that executes SQL queries for DB2 on z Systems. With the latest versions even more SQL is eligible for acceleration. It’s designed to be a very high performance appliance by nature. It provides transparent integration with DB2 on z Systems and should be thought of not as another database but an extension of DB2. Transparent integration means maintenance, administration and performance metrics are present in the current DB2 instruction facility. The user doesn’t need to learn a new interface and there are no application changes needed to utilize the appliance. In a nutshell, it has smoking fast queries.
Keyed to Analytics
IBM built the Analytics Accelerator as a piece of hardware that could be used in conjunction with the DB2 environment. It's keyed to analytics and offers a one-stop-shopping platform for very cost-effective data warehousing and data analytics. It's a marriage of data and speed of inquiry. In actuality, the marriage is between IBM z Systems high-end quality and IBM Netezza hardware accelerated technology. These two powereful solutions are combined to deliver state of the art performance at a lower cost of processing.
Working in conjunction with the z13 system, the Analytics Accelerator can bring north of 2,000 times improvement in very complex query performance compared to a DB2 platform executing the very same query. The High Performance Storage Saver feature lowers cost with processing, managing and storing both static and historical data.
This feature has the ability to keep static data in the DB2 environment versus on the z System platform. Doing this allows the data used for the queries to be moved directly to the appliance for processing, resulting in immediate savings. Even more savings can be realized by reducing the MIPS because the DB2 queries formerly running on z Systems can be passed directly to the hardware and processed there. Applications can keep the same format without performing any type of conversion as well. Additionally, data stored in the Analytics Accelerator possess exactly the same type of security that is used in DB2 for the z Systems platform.
So how does all of this work? The accelerator connects to the z/OS platform and DB2 via a private network. Using DB2 stored procedures and a GUI, the DB2 objects that are to be accelerated are defined and loaded to the Analytics Accelerator. The user sets up a few settings and choices and it’s ready to accelerate.
Once activated, the DB2 optimizer treats the appliance as a new access path and automatically offloads SQL for processing when it determines that it is more efficient to do so. Please keep in mind that not every query can be accelerated as some may have incompatible SQL code or the DB2 optimizer can determine that it is not the best choice. You also have the ability to tell the DB2 Optimizer to throw every eligible SQL query at the accelerator without consideration of cost.
Providing Business Value
The IBM DB2 Analytics Accelerator can provide deep business value in cost and speed. Additionally, transparency for applications and users, and also easy installation are huge wins for any organization. Based on an IBM Big Data & Analytics Hub blog post
“The potential benefits of Analytics Accelerator can be estimated before purchasing and installation. A study based on the information provided by the DB2 dynamic SQL statement cache can help you assess the SQL queries and the processor that would be offloaded if you deployed an accelerator. You can use a virtual accelerator in combination with EXPLAIN to understand how your queries would behave with a real accelerator. For example, the DSN_QUERYINFO_TABLE table provides reasons why ineligible queries are not eligible, as well as information about accelerated queries.
Business analytics are a must-have with today’s business culture. Data warehousing/business analytics are becoming mission-critical applications and require the same quality of service requirements as traditional online transaction processing workloads. Reliability, availability and security are what z Systems are all about. Today, another must-have is the ability to have the analytics workload coexist with traditional online processing in a mixed workload environment. This ability is key in determining your warehouse hosting platform, i.e., z Systems.
With the latest z13 mainframe release, the Analytics Accelerator is well positioned to help any major organization transform DB2 into a world-class transactional and analytic database engine with little effort. When thinking about total cost of ownership in your organization, that thought should include the benefits that the appliance can bring to the table.
Patrick Stanard is a z Systems Architect Manager for IBM. He is a 34 year professional in the industry spanning roles as a systems programmer, developer, manager, adjunct faculty member and director of operations. He has a Bachelor of Science in CIS from Saginaw Valley State University and an MBA from Michigan State University.