Co-Authors:
Satya Krishnaswamy, Director, HDM Development, IBM
Murali Madhanagopal, Software Architect, Intel
Jantz Tran, Software Engineer, Intel
Ryan Park, Sales Applications Engineer, Intel
The long-standing partnership between IBM and Intel has led to significant advancements in database performance over the past 25 years. Recently, this collaboration reached new heights with the announcement that Intel® Gaudi® 3 AI accelerators will be deployed as a service on IBM Cloud. This initiative represents yet another landmark in our joint venture. Moreover, IBM’s internal testing by IBM suggests the latest generation Intel® Xeon® Scalable processors from Intel, when integrated with Intel Software, could significantly enhance the performance of IBM watsonx.data and Db2. Businesses are looking to drive efficiency and scalability when managing critical workloads, whether on-premises or in the cloud. By harnessing this powerful combination, enterprises have the ability to efficiently handle complex tasks with speed and precision across various industries.
IBM Products
IBM watsonx.data
IBM watsonx.data is a new open, hybrid, and governed data lake house optimized for data, analytics, and AI workloads. The key highlights include:
- Driving analytics costs down with lower cost storage and analytic engines like Presto and Spark
- Using an open and flexible approach to provide a unified view of your data across hybrid cloud environments.
IBM Db2 Warehouse
IBM Db2 Warehouse is a cloud-native data warehouse optimized for critical, analytical workloads, featuring advanced MPP column-store technology and intelligent workload management for rapid data ingestion and high concurrency queries. It facilitates data sharing with support for open formats and integrates with watsonx.data for a unified analytics and AI view. Available as SaaS on AWS or IBM Cloud, or on-premise for hybrid data management architectures.
Intel Products & Technologies
4th & 5th Gen Intel® Xeon® Scalable Processors
Built on a shared architectural platform with the 4th Gen Intel® Xeon® scalable processor, the 5th Gen Intel® Xeon® scalable processor offers improved performance and performance per watt, TCO enhancements, and silicon-based security capabilities. In addition, it boosts performance for memory-bound and latency-sensitive workloads with faster memory and larger last-level cache compared to the previous generation.
Intel® Advanced Vector Extensions 512 (Intel® AVX-512)
The utilization of Single Instruction Multiple Data (SIMD) instructions like Intel AVX-512 can enable more data per instruction and complete multiple operations simultaneously. This enhanced processing power can with the ability to handle complex queries and analytics tasks with ease, delivering fast results and enhanced user experience. As organizations continue to demand quicker data processing times for real-time decision-making, watsonx.data and Db2's optimization leveraging Intel AVX-512 technology provides an edge by driving high performance leading to fast time to insight.
Improving Performance through Open-Source Software: Presto, Prestissimo, & Spark
Presto and Spark are the 2 primary query engines for watsonx.data and Db2.
- Presto query engine leverages the optimizations Intel has done to the Java Virtual Machine (JVM), vectorization, and garbage collection.
- Prestissimo is Presto’s next generation query engine leveraging C++ and SIMD instructions, built using the Velox library. The advantage of Prestissimo is huge performance boost and avoiding performance bottlenecks with JVM and garbage collection of Java. Intel is a member of Presto foundation along with IBM and has contributed AVX-512 optimizations to leverage vectorization for the query workers.
Workload Comparison: Performance improvements leveraging 5th Gen Intel® Xeon® Scalable processors
The results of the comparative analysis between 5th Gen Intel Xeon Scalable processors and their predecessors in watsonx.data and Db2 performance testing are truly groundbreaking. The IBM Big Data Insights (BDI) workload delivers a real-world simulation that mirrors complex retail environments, shedding light on the remarkable speed and agility of the latest generation Intel Xeon processors under high-stress conditions.
By subjecting the testing configuration to 16 concurrent users at a substantial 3TB scale factor, a robust evaluation framework was established to accurately measure and compare processor execution times. This demonstrates tangible advancements in performance metrics but also underscores the critical role played by cutting-edge technology in revolutionizing database operations. As businesses increasingly rely on data-driven insights for strategic decision-making, the significance of such comparative analyses becomes even more pronounced in guiding future hardware investments toward optimal efficiency and productivity gains.
Results:
IBM watsonx.data
We compared watsonx.data running on a single node using the BDI workload on four generations of Intel Xeon processors ranging from 2nd to 5th generation.
The 5th Gen Intel Xeon 8592+ processor stands out with up to 2.7X better query throughput than the 2nd Gen Intel Xeon scalable processor. Additionally, it boasts a 1.75X improvement over the 3rd Gen Intel Xeon 8380 processor and a 1.09X improvement over the 4th Gen Intel Xeon 8490H processor.
IBM Db2 Warehouse
Our comparison of Db2 Warehouse performance on a single node utilizing the BDI workload across four generations of Intel Xeon processors, ranging from the 2nd Gen to the latest 5th Gen Xeon processor, has yielded significant results. The graph showcases the Queries per Hour (QpH) achieved on each processor.
The 5th Gen Intel Xeon 8592+ processor stands out with up to 2.5X better query throughput than the 2nd Gen Intel Xeon 8280M processor. Additionally, it boasts a 1.64X improvement over the 3rd Gen Intel Xeon 8380 processor and a 1.15X improvement over the 4th Gen Intel Xeon 8490H processor.
Conclusion
As shown here, 5th Gen Intel Xeon scalable processors deliver superior performance. This workload takes advantage of the processor’s enhanced architecture, increased number of cores, and improved memory bandwidth. Customers will benefit significantly from faster response times for analytics queries, as well as higher throughput to support more concurrent users. This translates to cost savings and faster insights which users demand.
“For over a year, we have collaborated with Intel to optimize the price performance of Prestissimo. We’ve focused on the TPCDS analytical workloads benchmark. This optimization happens at every layer in the stack, from the query optimizer to the query engine and the storage tier. Our results to date show massive improvements compared to traditional open-source Presto.”
- Edward Calvesbert, IBM VP Product Management, watsonx.data
More
- Driving Superior watsonx Performance Through a Relationship with Intel
- IBM watsonx.data Accelerates GenAI Data Analysis
Configuration, Notices and Disclaimers
- For configurations please check 5th Generation Intel® Xeon® Scalable Processors - 1 | Performance Index
- Performance varies by use, configuration, and other factors. Learn more on the Performance Index site.
- Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.
- Your costs and results may vary.
- Intel technologies may require enabled hardware, software, or service activation.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.
IBM, the IBM logo, watsonx.data and Db2 are trademarks or registered trademarks of International Business Machines Corporation, in the United States and/or other countries. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on ibm.com/trademark.