R&D Thesis: Optimizing HTAP workloads using Matrix Multiplication

Introduction

IBM Research & Development is one of the largest private-sector research institutions in the world. Its portfolio ranges from hardware, firmware and operating systems for mainframes and storage technologies to Cloud, AI, Data Analytics, Automation, the Internet of Things and Quantum Computing. This also includes the operational maintenance of the first IBM quantum computer in Europe called “System One” and the development of critical technology components for the IBM Z Mainframe and Power Systems.

Your Role and Responsibilities

SAP HANA on IBM Power technology is used in business-critical solutions. Our development team working on the SAP campus in St. Leon Rot ensures that these solutions are running optimal on IBM Power architecture.

We are looking for a student who has interest in doing some research on a topic combining hardware accelerators and database technology. Background and topic of the master thesis:

In 2022, Zichun Huang and Shimin Chen made a proposal to use matrix multiplication for implementing a Join-Project database operator. To put it in a nutshell, the idea is to describe distinct values in a table as nodes in a graph. Relations in a table are described as Edge. Such tuples can be written as adjacence matrices. A Join-Project operation between two tables is then equivalent to the multiplication of each table’s adjacence matrix.

For a business application, several pieces of information are yet missing: (1) The paper evaluates their approach with graph datasets, not with OLAP datasets. They explicitly outrule TPC-H as usable dataset for their use case. The relevance for relational databases is therefore unclear. (2) Even though, each relation is stored as one bit, matrices may become very large. What is the `cost` to keep and maintain such matrices? (3) Currently, P10 MMA supports down to 4-bit integers, only. What matrix layout is reasonable if we interpreted the 16-Byte input as 1-bit integer type? (4) The original papers only evaluate their approach in isolated environments. How do we maintain such matrices in real databases with isolation levels and concurrent OLTP?

If you are interested to work in a dynamic and international team apply instantly.

Required Technical and Professional Expertise

• fluent in C/C++
• knowledge of parallel computing
• advanced knowledge about databases

Preferred Technical and Professional Expertise

• excellent communication skills
• fluent in English and German

IBM Systems helps IT leaders think differently about their infrastructure. IBM servers and storage are no longer inanimate – they can understand, reason, and learn so our clients can innovate while avoiding IT issues. Our systems power the world’s most important industries and our clients are the architects of the future. Join us to help build our leading-edge technology portfolio designed for cognitive business and optimized for cloud computing.

In a world where technology never stands still, we understand that, dedication to our clients success, innovation that matters, and trust and personal responsibility in all our relationships, lives in what we do as IBMers as we strive to be the catalyst that makes the world work better.

Being an IBMer means you’ll be able to learn and develop yourself and your career, you’ll be encouraged to be courageous and experiment everyday, all whilst having continuous trust and support in an environment where everyone can thrive whatever their personal or professional background.

Our IBMers are growth minded, always staying curious, open to feedback and learning new information and skills to constantly transform themselves and our company. They are trusted to provide on-going feedback to help other IBMers grow, as well as collaborate with colleagues keeping in mind a team focused approach to include different perspectives to drive exceptional outcomes for our customers. The courage our IBMers have to make critical decisions everyday is essential to IBM becoming the catalyst for progress, always embracing challenges with resources they have to hand, a can-do attitude and always striving for an outcome focused approach within everything that they do.

Are you ready to be an IBMer?

This job requires you to provide your COVID-19 vaccination status with supporting documentation, where legally permissible.

IBM’s greatest invention is the IBMer. We believe that through the application of intelligence, reason and science, we can improve business, society and the human condition, bringing the power of an open hybrid cloud and AI strategy to life for our clients and partners around the world.

Restlessly reinventing since 1911, we are not only one of the largest corporate organizations in the world, we’re also one of the biggest technology and consulting employers, with many of the Fortune 50 companies relying on the IBM Cloud to run their business.

At IBM, we pride ourselves on being an early adopter of artificial intelligence, quantum computing and blockchain. Now it’s time for you to join us on our journey to being a responsible technology innovator and a force for good in the world.

IBM ist es wichtig, Ihnen ein Arbeitsumfeld zu bieten, in dem Sie Ihr ganzes Potenzial voll entfalten können und die Möglichkeit haben, Ihren Arbeitstag flexibel zu gestalten. Wenn Sie Interesse daran haben, sprechen Sie uns bitte an, um herauszufinden, welche Gestaltungsmöglichkeiten im aktuellen Arbeitsumfeld bestehen.

IBM is committed to creating a diverse environment and is proud to be an equal opportunity employer. All qualified applicants will receive consideration for employment without regard to race, color, religion, gender, gender identity or expression, sexual orientation, national origin, genetics, pregnancy, disability, age, veteran status, or other characteristics. IBM is also committed to compliance with all fair employment practices regarding citizenship and immigration status.