Blogs

A Dynamic Duo for Analytics

DB2 with BLU Acceleration on Power Systems may transform the way organizations run analytics

Organizations as disparate as banking, healthcare, and law enforcement all share a common focus—analyzing information rapidly to provide enhanced delivery of services. For example, a bank needs split-second detection of fraudulent transactions so it can halt fraudsters before they do extensive damage. For law enforcement, real-time analysis of crime patterns can give police the ability to stop crime before it happens. And in healthcare, big data and analytics allow providers to improve services and reduce costs. Innovations today are assisting these sectors and many others to help meet and quickly solve time-driven data challenges.

In the world of big data and analytics, databases are workhorses used in analyzing data to generate insight. As the need for speed and agility increases, many organizations are switching to databases that use in-memory processing. As a result, IBM offers innovative BLU Acceleration technology that is built into the latest release of the IBM® DB2® 10.5 database for Linux, UNIX, and Windows and supported on both IBM Power® and x86 platforms. Organizations using DB2 with BLU Acceleration can experience significantly faster, cost-effective processing than ever before along with increased performance.

Innovative advances from IBM research and development labs enable BLU Acceleration to provide significant performance that offers speed-of-thought analytics through a simple load-and-go setup. DB2 with BLU Acceleration has been optimized for IBM Power Systems™ hardware to leverage the advanced design of the recent IBM POWER8™ processor-based technology. This first generation of systems built on the POWER8 technology is designed to deliver insight 50 times faster than traditional databases for highly demanding, data-hungry applications.*

The POWER8 processor is designed for big data, and it is capable of running increased concurrent queries in parallel quickly and across multiple cores by utilizing more threads per core. DB2 with BLU Acceleration is a next-generation database technology for in-memory computing. Organizations can analyze data rapidly and efficiently to uncover insights for helping grow revenue, reduce cost, and minimize risk.

Boosting performance

Traditional relational databases store data in rows to optimize read/write operations, which is a highly efficient way to perform transaction processing. But for analytics, which is a read-only operation, reading columns instead of rows is much more efficient. BLU Acceleration complements traditional row-based processing in DB2, and it speeds analytics and reporting by using next-generation, in-memory columnar technologies, which provide an extremely efficient way to scan and find relevant data. Placing the data in memory and reading just the specified columns can substantially improve performance. BLU Acceleration loads only the data needed and stores it in columnar format to maximize parallelism and throughput. By exploiting the higher number of threads per core, BLU Acceleration helps enhance performance for Power clients.

Unlike other in-memory database systems that require all data to be stored in memory to achieve desired performance levels, BLU Acceleration automatically optimizes movement of data from storage to system memory to processor memory. This optimization enables BLU Acceleration to maintain in-memory performance even when active data sets are larger than system memory capacity. Needed information can be obtained at high speed and volume.

DB2 was designed to maximize performance for both analytics and transaction processing. It can support BLU Acceleration for analytics at the same time as transactions are being processed. BLU Acceleration is thrifty when it comes to using memory because of the following advances that enable BLU Acceleration to put only one-tenth of data in memory:

  • Dynamic in-memory data storage that exploits in-memory columnar processing without the limitations or cost of in-memory-only systems
  • Actionable compression application that preserves order so that the data can be used without first decompressing it
  • Parallel vector processing that is designed to deliver multi-core and Single Instruction Multiple Data (SIMD) parallelism; this technology allows work to be spread across multiple processor cores at the same time that POWER8 or Intel processors are working on multiple data elements in each processing cycle
  • Data skipping that uses intelligence to ignore irrelevant data during processing

These advances yield fast insights from high volumes of data. As a result, organizations can make better decisions and better serve their customers than they can with traditional approaches to analytics and transaction processing, while reducing risk and improving operational efficiency.

Organizations that deploy POWER8 can capitalize on DB2 by automatically harnessing the platform’s hardware parallelism without requiring applications to be rewritten. DB2 also automatically uses the large page sizes available with POWER8 and aligns resource objects with the system architecture. This DB2 and Power Systems synergy with POWER8 helps simplify deployment of database applications, achieve cost-effective large-scale processing, and consolidate workloads.

Solving real-world challenges

The combination of BLU Acceleration with Power Systems enables organizations to put data to work and generate insights quickly. For example, a large health insurance company with the capability to analyze high volumes of data faster than was previously possible can offer enhanced customer service to its clients that helps reduce healthcare expenses and provide improved medical management. In this way, the healthcare organization can put data to work to solve real business challenges.

Because BLU Acceleration is fully integrated with DB2, implementation doesn’t require SQL or schema changes. Automation, optimization, and self-tuning make BLU Acceleration easy to deploy. To make deployment as simple as possible, IBM will also preinstall and optimize DB2 with BLU Acceleration on a new POWER8 server. This service can speed up time of deployment and let the client organization focus on addressing real-world challenges.

BLU Acceleration deployments are designed to help organizations use the technology cost-effectively to uncover insights that can grow revenue. For example, they can reduce runtimes of specific queries from hours down to minutes. Further, the compression technologies in DB2 combined with the efficiencies of POWER8 help keep the hardware footprint small and minimize power, cooling, and data center space costs.

DB2 can also take advantage of the IBM PowerVM® virtualization technology available with Power Systems. By virtualizing processor, memory, and I/O resources and enabling partitioning of each processing core, PowerVM allows organizations to consolidate workloads onto fewer physical servers. This consolidation helps reduce hardware and energy costs while creating a highly flexible IT infrastructure. And Power Systems provides virtualization that empowers organizations to run data warehouse and analytics applications on the same server, which helps reduce the number of servers needed in the data center.

Applying robust analytics

Organizations can continue to rely on DB2 on Power Systems for traditional transaction processing, leveraging the resiliency, availability, and security that the platform provides. Although BLU Acceleration doesn’t currently support transaction processing, DB2 can run transactions using traditional row-based processing while using BLU Acceleration’s in-memory, columnar processing to run analytics queries.

Data demands and analytics requirements are growing daily, and many organizations are looking for robust yet simple solutions. For organizations seeking to benefit from big data and analytics, deploying Power Systems and DB2 with BLU Acceleration offers an impressive combination for deriving key insights that help them successfully remain efficient and competitive.

Please share any thoughts or questions in the comments.

* Based on internal tests by IBM at IBM performance labs in April 2014. The testing compared an IBM DB2 with BLU Acceleration system against a comparably tuned competitor configuration executing a materially identical 2.6 TB operational analytics workload in a controlled laboratory environment. The test measured 60 concurrent end-user report throughputs executing identical IBM Cognos® report workloads. The competitor configuration consisted of an HP DL380p, 24 cores, 256 GB RAM, a traditional database, SuSE Linux 11SP3 (database) and HP DL380p, 16 cores, 384 GB RAM, Cognos 10.2.1.1, and SuSE Linux 11SP3 (Cognos). The IBM configuration consisted of an IBM POWER8, 24 cores, 256 GB RAM, DB2 10.5, IBM AIX 7.1 TL2 (database) and IBM POWER8, 16 cores, 384 GB RAM, Cognos 10.2.1.1, and SuSE Linux 11SP3 (Cognos). Results may not be typical and will vary based on actual workload, configuration, applications, queries, and other variables in a production environment.