Each quarter, Red Hat Research Quarterly highlights new and ongoing research collaborations from around the world.
This quarter we highlight a few collaborative projects from the United States at Boston University, Emory University, Tufts University, Northeastern University, and Columbia University, and in Germany at TU München. Contact academic@redhat.com for more information on any project described here or visit the Red Hat Research Project Directory.
PROJECT: Characterizing microservice architectures
ACADEMIC INVESTIGATORS: Prof. Raja Sambasivan (lead) and PhD students Darby Huye and Lan Liu (Tufts); Prof. Avani Wildani and PhD student Vishwanath Seshagiri (Emory)
Red Hat investigators: Not specified (industry participants are anonymous due to the nature of this research)
Prof. Raja Sambasivan and his students collaborated with Prof. Avani Wildani of Emory University and her students to complete a user study with microservice developers. The study aims to identify the variety of industrial microservice architectures in use by developers in industry and compare those architectures to those available in current open source testbeds such as DeathStarBench, TrainTicket, and BookInfo. Investigators used codebase analyses, literature reviews, and developer interviews to probe industrial microservice designs in detail.
The researchers found that industrial architectures vary greatly from testbeds and that this may invalidate research assumptions for work using existing testbeds. The teams hope to broaden open source testbeds and to continue identifying and classifying microservice architectures in future work. This work will also benefit their previous work on tools and data collection for large-scale traces in realistic distributed system computing topologies and their work on open telemetry with Red Hat and OpenInfra Labs. Their article “SoK: Identifying mismatches between microservice testbeds and industrial perceptions of microservices” was published in the Journal of Systems Research 2:1, June 2022.
PROJECT: LEGO for 6G: A modular open access end-to-end network architecture to enable B5G/6G research
ACADEMIC INVESTIGATORS: Abhimanyu Gosain, Senior Director, Institute of Wireless Internet of Things; Prof. Tommaso Melodia, IWIOT Institute Director; Michele Polese, Principal Research Scientist (Northeastern)
Red Hat investigator: Dan Winship
The IoT team’s work evolved from 5G to 6G since the last research update, suggesting a roadmap for leveraging public-private partnership research testbeds to influence the 6G vision and accelerate the entire lifecycle of research and development, manufacturing, standardization, and market readiness for 6G. Following a recent press release announcing the formation of the Open6G Cooperative Research and Development Center at Northeastern, the group is exploring topics such as spectrum access and exploitation, Open Radio Access Network (O-RAN) architectures, AI/ML for inference and control, and mmWave and Terahertz systems.
Traditional vendors treat the Radio Unit (RU) and Distribution Unit (DU) of a Radio Access Network (RAN) together as a black box. The international testbed where Gosain’s group and collaborators work, which is now running OpenShift for experiment and node deployments, allows separate experimentation with RU and DU components. This unlocks a lot of interesting possibilities for optimizing and controlling RANs. Abhimanyu Gosain presented this work at the September Research Day event in Brno; slides from the presentation and a video recording of the session are available.
PROJECT: Near-data data transformation
ACADEMIC INVESTIGATORS: Prof. Manos Athanassoulis (PI), Prof. Renato Mancuso (Co-PI), and PhD students Shahin Roozkhosh and Tarikul Islam Papon (BU); Denis Hoorneart, PhD Student (TU München)
Red Hat investigators: Uli Drepper and Ahmed Sanaullah
This team re-examined their previously implemented software-hardware co-design approach, which pushed data transformation closer to memory from a real-time systems perspective. By doing realistic prototyping with CPU + FPGA platforms, the team has been able to allow efficient and cache-friendly access to large data objects by moving only relevant data items from target memory. This compressed the working-set size, and therefore the cache footprint, and reorganized complex memory access patterns into sequential access with predictable patterns.
The team dubbed this approach Programmable Logic In-the-Middle (PLIM), presenting it as an important part of an Open Hardware Initiative Series at DevConf.US 2022, in both a talk and a Q&A panel discussion. Prof. Mancuso also presented related research ideas at the October virtual Red Hat Research Days event “Can we control time? Toward knowledge-driven system management to control timeliness.” A video recording can be found in the Past Events section of the Red Hat Research website.
PROJECT: Automated detection of memory safety vulnerabilities in Rust
ACADEMIC INVESTIGATORS: Prof. Baishaki Ray and PhD student Vikram Nitin (Columbia)
Red Hat investigators: The (Anne) Mulhern and Sanjay Arora
When Rust code is compiled, it goes through a High-level Intermediate Representation (HIR) and a Mid-level Intermediate Representation (MIR). This project team has implemented a hybrid analysis that combines information from the HIR and the MIR to detect code patterns that may cause memory safety violations, such as incompatible lifetime dataflow conditions. The team has been using the system, implemented as a subroutine within the RUDRA project, to examine Rust code in open source repositories.
The team is currently analyzing results from this process and expects to make their code available in a publication soon. For more on the team’s work, see their virtual Research Day event. For more information about RUDRA, see Yechan Bae et al., “RUDRA: finding memory safety bugs in Rust at the ecosystem scale,” in Proceedings of the ACM SIGOPS 28th Symposium on Operating Systems Principles. (2021).