Marius Pirvu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Marius Pirvu has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Harpreet Kaur
Department of Electrical Engineering and IT, Punjab Agricultural University, Ludhiana, Punjab, India
2 shared publications - Kenneth Blair Kent
University of New Brunswick
2 shared publications - Scott Ryan Young
Faculty of Computer Science, University of New Brunswick, Fredericton, New Brunswick, Canada
2 shared publications - Vijay Sundaresan
IBM Canada Lab, Markham, Canada
2 shared publications - Daryl Maier
IBM Canada Lab, Markham, Canada
1 shared publication Ian Gartley
IBM Canada Ltd
1 shared publication
Latest Publications
- Open Access
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2054-2072
In-Production Characterization of an Open Source Serverless Platform and New Scaling Strategies
Serverless computing has become more popular and evolved to support more complex tasks than the original Function as a Service (FaaS) model. The design of serverless systems has advanced to accommodate application demands and offer flexibility. Careful characterization of modern serverless systems and understanding of current gaps are warranted. Publicly available datasets on workloads in select production serverless systems do not fully represent all offerings or capture traces at the required time resolution to identify changes in application-level request-response patterns. We characterize - and make available - production traces from a major public serverless provider with over 1.9 billion invocations spanning over two months. Our dataset is the first to characterize an open-source platform with large-scale trace data, millisecond-scale arrival times, and user configurations for pod concurrency and minimum pod scaling. Using this dataset, we provide new insights for optimizing serverless platforms. With our insights, we design and implement FeMux, a serverless lifetime management system. FeMux multiplexes lightweight forecasters and uses a Representative Unified Metric (RUM) to decouple metrics from serverless platform implementations. This allows providers like us to update metrics flexibly or support multiple system objectives simultaneously. We prototype FeMux on the Knative platform and evaluate its benefits.
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2054-2072
Open AccessIn-Production Characterization of an Open Source Serverless Platform and New Scaling Strategies
Serverless computing has become more popular and evolved to support more complex tasks than the original Function as a Service (FaaS) model. The design of serverless systems has advanced to accommodate application demands and offer flexibility. Careful characterization of modern serverless systems and understanding of current gaps are warranted. Publicly available datasets on workloads in select production serverless systems do not fully represent all offerings or capture traces at the required time resolution to identify changes in application-level request-response patterns. We characterize - and make available - production traces from a major public serverless provider with over 1.9 billion invocations spanning over two months. Our dataset is the first to characterize an open-source platform with large-scale trace data, millisecond-scale arrival times, and user configurations for pod concurrency and minimum pod scaling. Using this dataset, we provide new insights for optimizing serverless platforms. With our insights, we design and implement FeMux, a serverless lifetime management system. FeMux multiplexes lightweight forecasters and uses a Representative Unified Metric (RUM) to decouple metrics from serverless platform implementations. This allows providers like us to update metrics flexibly or support multiple system objectives simultaneously. We prototype FeMux on the Knative platform and evaluate its benefits.
- Open Access
- 28 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 63-69
Early Detection of Unsupported Compilations during Prototyping in a Template-based Just-in-Time Compiler
- 28 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 63-69
Open AccessEarly Detection of Unsupported Compilations during Prototyping in a Template-based Just-in-Time Compiler
- Open Access
- 8 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 1114-1121
Lightweight Compilation of Method Invocation Bytecodes in Java
Method invocation bytecodes are a fundamental aspect of class methods and object interactions in object-oriented programming languages. However, their traditional interpretation and compilation techniques often introduce significant overhead, impacting the execution efficiency of applications. Template-based Just-In-Time compilation is a lightweight approach in Eclipse OpenJ9 that leverages pre-defined templates to convert frequently executed code fragments into machine code, significantly cutting down startup times. This paper extends the template-based approach to compilation of method invocation bytecodes in Java, aimed at reducing the overhead associated with compiling methods and improving the overall runtime performance. To demonstrate the effectiveness of the proposed approach, a prototype implementation is developed and evaluated against the industry-standard DaCapo benchmark suite. The evaluation reveals promising results, showcasing notable performance gains in terms of reduced memory footprint, faster startup times, and improved overall execution efficiency. This research contributes to the ongoing efforts in enhancing the performance of Java bytecodes' execution and eventually the Java applications. The insights gained from this study are expected to inspire further research in the domain of bytecode compilation, ultimately benefiting the broader language development community.
- 8 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 1114-1121
Open AccessLightweight Compilation of Method Invocation Bytecodes in Java
Method invocation bytecodes are a fundamental aspect of class methods and object interactions in object-oriented programming languages. However, their traditional interpretation and compilation techniques often introduce significant overhead, impacting the execution efficiency of applications. Template-based Just-In-Time compilation is a lightweight approach in Eclipse OpenJ9 that leverages pre-defined templates to convert frequently executed code fragments into machine code, significantly cutting down startup times. This paper extends the template-based approach to compilation of method invocation bytecodes in Java, aimed at reducing the overhead associated with compiling methods and improving the overall runtime performance. To demonstrate the effectiveness of the proposed approach, a prototype implementation is developed and evaluated against the industry-standard DaCapo benchmark suite. The evaluation reveals promising results, showcasing notable performance gains in terms of reduced memory footprint, faster startup times, and improved overall execution efficiency. This research contributes to the ongoing efforts in enhancing the performance of Java bytecodes' execution and eventually the Java applications. The insights gained from this study are expected to inspire further research in the domain of bytecode compilation, ultimately benefiting the broader language development community.
- 1 February 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Experiences in designing a robust and scalable interpreter profiling framework
Profile directed feedback (PDF) is a well known technique used to drive many compiler optimizations like basic block ordering and guarded devirtualization. These optimizations are particularly crucial in order to achieve good throughput performance in JEE applications that have a large code footprint. To effectively apply optimizations that rely on profiling information, a just-in-time (JIT) compiler must have access to profiling information that is accurate. One common source of profiling information in a Java virtual machine (JVM) is the interpreter. Typically methods are interpreted as a program ramps up, during which profiling information can be collected. However, obtaining useful and accurate information for large enterprise-class applications can be a challenge because of the memory and performance overhead associated with collecting and processing the large volumes of profiling data that is generated. This paper describes the challenges in maintaining the balance between throughput performance and profiling overhead in a production JIT compiler that is used by the IBM JDK. The scope of the performance overhead in terms of throughput, memory footprint and startup speed for large JEE class applications is introduced and various engineering solutions that were tried are detailed and compared in terms of experimental results. We found that the throughput improvement due to interpreter profiling (IP) can be as high as 58%, whereas the overhead measured in terms of application startup time could cost up to 57%. Our solutions to reducing profiling overhead managed to reduce the startup cost to only a few percent while maintaining the full throughput benefit. By discussing these approaches, this paper offers a balanced and practical overview on how to make PDF work well for enterprise-class applications in a production JIT compiler.
- 1 February 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Experiences in designing a robust and scalable interpreter profiling framework
Profile directed feedback (PDF) is a well known technique used to drive many compiler optimizations like basic block ordering and guarded devirtualization. These optimizations are particularly crucial in order to achieve good throughput performance in JEE applications that have a large code footprint. To effectively apply optimizations that rely on profiling information, a just-in-time (JIT) compiler must have access to profiling information that is accurate. One common source of profiling information in a Java virtual machine (JVM) is the interpreter. Typically methods are interpreted as a program ramps up, during which profiling information can be collected. However, obtaining useful and accurate information for large enterprise-class applications can be a challenge because of the memory and performance overhead associated with collecting and processing the large volumes of profiling data that is generated. This paper describes the challenges in maintaining the balance between throughput performance and profiling overhead in a production JIT compiler that is used by the IBM JDK. The scope of the performance overhead in terms of throughput, memory footprint and startup speed for large JEE class applications is introduced and various engineering solutions that were tried are detailed and compared in terms of experimental results. We found that the throughput improvement due to interpreter profiling (IP) can be as high as 58%, whereas the overhead measured in terms of application startup time could cost up to 57%. Our solutions to reducing profiling overhead managed to reduce the startup cost to only a few percent while maintaining the full throughput benefit. By discussing these approaches, this paper offers a balanced and practical overview on how to make PDF work well for enterprise-class applications in a production JIT compiler.