Master’s Student Marijn Vollaard Shines with Study on Hardware Dimensioning for Microservice Applications in Cyber-Physical Systems

Our master’s student, Marijn Vollaard, has achieved a significant milestone by completing and presenting his literature study titled “Hardware Dimensioning for Microservice Applications in Cyber-Physical Systems: Current Directions and Challenges” The study addresses the challenge of dimensioning the number of compute nodes required to meet the performance demands of microservice-based applications in cyber-physical systems. It thoroughly reviews an extensive body of literature on application and system profiling, performance prediction, and design-space exploration to establish the current state of knowledge in this field. The survey culminates in a discussion about how the surveyed literature applies to microservice applications, the cyber-physical systems context, and the problem of hardware dimensioning. Overall, this is a nice piece of work with a lot of references presented in a systematic way. Congratulations to Marijn for his great effort!”

Bridging the Gap: Rethinking Real-Time Systems for Industry Success and Model-Driven Performance Engineering

The real-time systems community is shrinking and needs to bridge the gap between academic research an industry practice. In my pitch at ECRTS, I shared our view on model-driven system performance engineering for cyber-physical systems and encouraged the community increase its scope and take a broader responsibility for timing-related issues in systems to achieve those goals. This means working in more of the focus areas that we have identified in our vision and validated with our industry partners, but also reconsidering some directions in areas where work is already taking place. This means less focus on hard real-time requirements and formal methods and more focus on:

• system-level KPIs instead of meeting deadlines in subsystems
• soft real-time requirements
• timing requirements beyond software
• system performance modelling, model calibration, and model learning
• data-driven performance analysis, optimization, verification, and diagnostics, e.g. using traces

I encouraged the community to have a look at our vision for model-based system performance engineering for industrial cyber-physical systems and asked to think about how they could contribute through their current and future work.

Please have a look at our vision here.

Thanks to Bram van der Sanden, Kuan-Hsun Chen, Mitra Nasri, Geoffrey Nelissen, and Twan Basten for their help preparing the pitch.

Keynote Address Explores Performance Engineering in Cloud-Connected Cyber-Physical Systems

I had the honor of being invited as keynote speaker at RT-Cloud 2023. The keynote discussed the increasing complexity of cyber-physical systems (CPS) in the Dutch high-tech systems industry and a gradual transition towards microservice architectures and cloud-connected systems. This transition challenges our experience with performance engineering in the CPS domain, as we adapt our methods to embrace new tools and technologies. To make the presentation concrete, I discussed two projects that I am currently working on, a project on performance verification of microservice architectures together with Thales, and a project about performance engineering and service continuity in the compute continuum, together with Philips and TU/e and other TRANSACT partners. I would like to thank Johan Eker and Luca Abeni for the invitation and all participants for their attention and questions.

Inaugural Lecture Explores Managing Complexity of High-Tech Systems

Today, I finally gave my inaugural lecture “Managing Complexity in High-tech Systems” to celebrate my appointment as Endowed Professor at the University of Amsterdam, which happened back in 2019.

The academic ceremony started at 16:00 with a small reception for fellow professors and members of the curatorium. Together, this group walked in a procession into the beautiful auditorium of the University of Amsterdam, where an audience of colleagues, family, and friends, where waiting in anticipation. The lecture discussed the challenge of increasing complexity in the high-tech equipment industry and how new (model-based) development methodologies leveraging abstraction, boundedness, and composition, are required to address it. I argued that the required innovation should come from collaboration in an innovation chain, where universities, applied research organizations, and industry work together in strategic partnerships. The presentation was concluded with a number of concrete examples of what this collaboration could look like, based on examples from my education and research at TNO and the University of Amsterdam. The inaugural lecture was followed by a reception full of networking and congratulations. I would like to thank everybody that showed up for the event, physically and online. Together, we created a memory that I will treasure for a lifetime.

If you did not manage to attend the lecture, or see it online, there is a recording available. Pop some popcorn, take a seat, and click the link below:

https://webcolleges.uva.nl/Mediasite/Play/99497b81432a49acb57f0ae7a32050d11d

TNO-ESI Cloud Continuum Workshop Connects Researchers and Promotes Collaboration in the Netherlands

The TNO-ESI Cloud Continuum workshop, an informal hybrid event that attracted just over twenty participants, took place at ESI on February 21. The goals of this workshop were to: 1) connect applied and academic researchers in the area of cloud continuum in the Netherlands, 2) disseminate research results from ongoing research projects, and 3) identify possibilities for collaboration. Benny Akesson, the organizer of the event, opened the workshop by presenting some drivers for cloud adoption/integration in the high-tech industry, as well as the work done by ESI in the ArchViews and TRANSACT projects related to performance observability. This was followed by four invited speakers from Eindhoven University of Technology and Vrije Universiteit Amsterdam. The topics of the presentations ranged from reference architectures for the cloud continuum, root-cause analysis in the continuum, modelling and calibration of cyber-physical systems deployed in the continuum, to performance variability of cloud/edge systems. All in all, it was a nice and successful event that showcased parts of the body of work currently going on in this exciting area. Thank you Matthijs Jansen, Jeroen Voeten, Mahtab Modaber, and Panagiotis Giannakopoulos for your presentations.

Design Methodologies for Cyber-physical Systems

In this short two minute presentation, I introduce myself and my fundamental and academic research into design methodologies for cyber-physical systems. I sketch a high-level view of the problem and outline a direction based on model-based engineering in which my previous work into domain-specific languages and analysis non-functional behavior fits. For a more elaborate description of my research, please have a look at my research page.

Back from MODELS 2019

After six days in Munich I have now left the MODELS 2019 conference. It has been an intense couple of days with three days of workshops and tutorials, and three days of main conference. Both the technical and social aspects of the conference were exceptionally well-organized, so kudos to the men and women who worked hard to make that happen.

The four main highlights at the conference for me were:
1. Presenting our paper “Towards Continuous Evolution through Automatic Detection and Correction of Service Incompatibilities” at the MODCOMP workshop. Discussions with conference participants about Petri Net transformations have given inspiration for how to formally work with more complex service behaviors than we do in our work on service-oriented architectures today.

2. A tutorial on StateCharts that improved my understanding of a model-of-computation I will be teaching at the University of Amsterdam in the near future. Thanks to Simon van Mierlo, Hans Vangheluwe, and Axel Terfloth for organizing this tutorial and for sharing their excellent material.

3. Meeting and discussing with representatives from BMW, Daimler, MAN, Continental, TTTech, and other automotive companies and hear more about automotive trends towards centralization of computation, first through domain controllers and then further towards integration of domains in automotive “supercomputers”. It was also interesting to see that the automotive industry is showing interest in service-oriented architectures as a paradigm for their platforms. In fact, a paper entitled “Model-Based Resource Analysis and Synthesis of Service-Oriented Automotive Software Architectures” from BMW got the Best Paper Award on the Practice and Innovation track for work in this direction. This confirms our belief that our current applied research on service-oriented architectures in the defense domain can be generalized to other domains.

4. Meeting and talking to people from both Flanders Make and CETIC, which are the Flemish and Wallonian equivalents of ESI (TNO). It was interesting to talk to them and learn about how what we do is similar and different, both in terms of technical scope and business models.

I hope to return to the MODELS conference again next year to present more of our work and have another opportunity to discuss with and learn from top academics and industrialists in the area of model-based engineering.

Journal Article Presented at ECRTS 2019

Today, Ali presented our Real-time Systems article “Uneven Memory Regulation for Scheduling IMA Applications on Multi-core Platforms” in the Journal-to-conference (J2C) session at ECRTS.

This article addresses the problem of resource sharing in mixed-criticality systems through temporal isolation by extending the state-of-the-art Single-Core Equivalence (SCE) framework in three ways: 1) we extend the theoretical toolkit for the SCE framework by considering EDF and server-based scheduling, instead of partitioned fixed-priority scheduling, 2) we support uneven memory access budgets on a per-server basis, rather than just on a per-core basis, and 3) we formulate an Integer-Linear Programming Model (ILP) guaranteed to find a feasible mapping of a given set of servers to processors, including their execution time and memory access budgets, if such a mapping exists. Our experiments with synthetic task sets confirm that considerable improvement in schedulability can result from the use of per-server memory access budgets under the SCE framework.

Overall, I greatly appreciate that key conferences in the real-time community are starting to allow journal articles to be presented. This increases the exposure of these works that are often longer and better edited. It is also helpful for researchers at the institutes where conference publications are not considered a relevant KPI. You can argue the validity of this reasoning in areas of computer science where conferences are highly competitive with 20-30% acceptance rates, but it is reality for some researchers. An interesting thing with the MODELS conference is that they collaborate with the SOSYM journal such that some accepted articles in the journal gets a full slot at the conference. This is a nice way to highlight good articles and to appreciate the work done by both authors and reviewers.

Paper Accepted at EMSOFT 2019

Our collaboration with CISTER has been extremely fruitful this year, as yet another paper in our research line on mixed-criticality scheduling has been accepted. This latest paper is entitled “Techniques and Analysis for Mixed-criticality Scheduling with Mode-dependent Server Execution Budgets” and has been accepted at the International Conference on Embedded Software (EMSOFT).

The goal of this work is, like many other in this research line, is to reduce cost of mixed-criticality systems. This time, we achieve this by addressing the limitation that a server only has a single execution budget in all modes, despite that their computational requirements may vary significantly. More specifically, the three main contributions of the paper are: 1) a scheduling arrangement for uni-processor systems employing fixed-priority scheduling within periodic servers, whose budgets are dynamically adjusted at run-time in the event of a mode change, 2) a new schedulability analysis for such systems, and 3) heuristic algorithms for assigning budgets to servers in different modes and ordering the execution of the servers. Experiments with synthetic task sets demonstrate considerable improvements (up to 52.8%)

Invited Presentation at CMAS 2015

I have recently accepted an invitation to speak at the First TCRTS Workshop on Certifiable Multicore Avionics Systems (CMAS), which takes place on April 13 and is co-located with RTAS 2015 in Seattle. The presentation is made in collaboration with Jan Nowotsch at Airbus Group Innovations, where I was a Visiting Researcher during two months last year. The title of the presentation is Towards Certifiable Resource Sharing in Safety-Critical Multi-Core Real-Time Systems and discusses current problems and state-of-the-art methods for resource sharing in real-time multi-core platforms. The abstract of the presentation is found below:

The proliferation of multi-core platforms has had great impact on embedded computing. Multiple cores exploiting task-level parallelism offer performance far beyond what is possible with a single core, while staying within an acceptable power envelope. Since resources, such as interconnect and memories, are often shared between cores, the platforms have also become increasingly cost efficient. However, resource sharing results in interference between concurrently executing applications, which causes problems in real-time systems where such interference must be either bounded or completely eliminated. As a result, safety-critical systems, for example in the avionics domain, have not yet been able to capitalize on the benefits of multi-core platforms due to stringent certification requirements.

This presentation discusses the state-of-the-art in resource sharing in multi-core systems and its application to safety-critical real-time systems. First, a survey of efforts to build time-predictable resources, such as interconnects and memory controllers, is provided. Then, software-based interference mitigation mechanisms and analyses for these resources in commercial-of-the-shelf platforms are discussed. This is followed by an overview of the approach proposed by Airbus Group Innovations to manage interference and compute worst-case execution times of applications running on a Freescale P4080 multi-core platform. The presentation is concluded by highlighting open issues and future directions towards certifiable resource sharing in safety-critical multi-core real-time systems.

Update: The slides are available here.