SFSCON24 - Michal Skipala & Bruno Rossi - Monolith Splitter
0 likes13 views
The document discusses the development of Monolith Splitter, a tool designed to facilitate the migration of monolithic applications to a microservices architecture, focusing on the challenges and complexities involved in this transition. It highlights the tool’s ability to analyze execution traces and provide clustering recommendations to optimize migration processes, while also detailing its tested implementations and planned enhancements. Additionally, it addresses the limitations and challenges faced in achieving effective migrations, emphasizing the ongoing need for research and collaboration in this area.
![Mono2Micro
● Spatio-temporal decomposition
● Business cases
● Runtime call relations
○
○
● Direct Call Relations, Direct Call Pattern
● Indirect Call Relations, Indirect Call Pattern
● Computation of Similarity
● Hierarchical Clustering
[1] Mono2Micro: A Practical and Effective Tool for Decomposing Monolithic Java
Applications to Microservices
https://www.researchgate.net/publication/354057927_Mono2Micro_a_practical_and_
effective_tool_for_decomposing_monolithic_Java_applications_to_microservices
[1]](https://image.slidesharecdn.com/1440-michalskipalabrunorossi-monolithsplitter-241118080743-6179440a/85/SFSCON24-Michal-Skipala-Bruno-Rossi-Monolith-Splitter-7-320.jpg)
![FoSCI
● Functional Atom Generation (coherent, minimal functional unit)
● Assignment of Functional Atoms into Service Candidates
● Non-dominated Sorting Genetic Algorithm-II (NSGA-II) for optimization
● Structural and Conceptual Intra-Connectivity & Inter-Connectivity
[2] Service Candidate Identification from Monolithic Systems Based on Execution
Traces https://ieeexplore.ieee.org/document/8686152
[2]](https://image.slidesharecdn.com/1440-michalskipalabrunorossi-monolithsplitter-241118080743-6179440a/85/SFSCON24-Michal-Skipala-Bruno-Rossi-Monolith-Splitter-8-320.jpg)
![Monolith Splitter Implementation & Integration
● Two reimplemented approaches
● Implementation based on research articles:
[1] Mono2Micro: A Practical and Effective
Tool for Decomposing Monolithic Java
Applications to Microservices
[2] Service Candidate Identification from
Monolithic Systems based on Execution
Traces
[1] Mono2Micro: A Practical and Effective Tool for Decomposing Monolithic Java
Applications to Microservices
https://www.researchgate.net/publication/354057927_Mono2Micro_a_practical_and_
effective_tool_for_decomposing_monolithic_Java_applications_to_microservices
[2] Service Candidate Identification from
Monolithic Systems Based on Execution Traces
https://ieeexplore.ieee.org/document/8686152](https://image.slidesharecdn.com/1440-michalskipalabrunorossi-monolithsplitter-241118080743-6179440a/85/SFSCON24-Michal-Skipala-Bruno-Rossi-Monolith-Splitter-9-320.jpg)
![Monolith Splitter Test Results
● Currently tested on the JPetStore [3] and AcmeAir [4] applications
● Execution traces sourced from publicly available Mono2Micro datasets on
GitHub [5]
● Future testing planned on widely-used, monolithic open-source projects
written in Java
[3] JPetStore https://github.com/KimJongSung/jPetStore
[4] AcmeAir https://github.com/acmeair/acmeair
[5] Mono2Micro-FSE-2021 https://github.com/kaliaanup/Mono2Micro-FSE-2021](https://image.slidesharecdn.com/1440-michalskipalabrunorossi-monolithsplitter-241118080743-6179440a/85/SFSCON24-Michal-Skipala-Bruno-Rossi-Monolith-Splitter-10-320.jpg)
SFSCON24 - Michal Skipala & Bruno Rossi - Monolith Splitter
- 1. Monolith Splitter SFSCON 2024 Support for Migration of a Monolithic System to Microservices
- 2. Speakers Bruno Rossi ● Assistant Professor at the at the Faculty of Informatics, Masaryk University, Brno, Czech Republic ● Research interests in software evolution, open source software systems, and cyber-physical systems ● Integrating open source software in the courses taught at Masaryk University Michal Skipala ● Software Engineer at T-Mobile Czech Republic ● Student at Masaryk University, currently working on Diploma Thesis ● Specializes in Fullstack Web Development in recent technologies ● Proficient in Java and ReactJS
- 3. Context ● Migration from monolithic systems is of interest for many companies running on legacy systems, but… ○ Complexity of migration to microservices architecture ○ Numerous ways to split a monolith into a microservices system ○ Measuring the quality of the final migration is challenging ● Research papers suggesting ways to optimize the process often lack implementation
- 4. What is Monolith Splitter? ● A tool in development that suggests approaches for migrating monolithic applications to microservices ● Accepts execution traces and Git log files as input ● Provides recommendations for clustering classes from execution traces into microservice-aligned groups ● Tech Stack: Java 21 + Spring Boot 3.2, PostgreSQL, TypeScript, ReactJS
- 5. Benefits ● Monolith Splitter provides support for software architects aiming to transition applications to a microservices architecture ● Algorithm implementations will be openly available, enabling practitioners to leverage the tool across various projects ● Accelerates the modernization of legacy applications by providing optimized strategies for monolith decomposition
- 6. Example Input & Output I: Execution traces O: Clusters
- 7. Mono2Micro ● Spatio-temporal decomposition ● Business cases ● Runtime call relations ○ ○ ● Direct Call Relations, Direct Call Pattern ● Indirect Call Relations, Indirect Call Pattern ● Computation of Similarity ● Hierarchical Clustering [1] Mono2Micro: A Practical and Effective Tool for Decomposing Monolithic Java Applications to Microservices https://www.researchgate.net/publication/354057927_Mono2Micro_a_practical_and_ effective_tool_for_decomposing_monolithic_Java_applications_to_microservices [1]
- 8. FoSCI ● Functional Atom Generation (coherent, minimal functional unit) ● Assignment of Functional Atoms into Service Candidates ● Non-dominated Sorting Genetic Algorithm-II (NSGA-II) for optimization ● Structural and Conceptual Intra-Connectivity & Inter-Connectivity [2] Service Candidate Identification from Monolithic Systems Based on Execution Traces https://ieeexplore.ieee.org/document/8686152 [2]
- 9. Monolith Splitter Implementation & Integration ● Two reimplemented approaches ● Implementation based on research articles: [1] Mono2Micro: A Practical and Effective Tool for Decomposing Monolithic Java Applications to Microservices [2] Service Candidate Identification from Monolithic Systems based on Execution Traces [1] Mono2Micro: A Practical and Effective Tool for Decomposing Monolithic Java Applications to Microservices https://www.researchgate.net/publication/354057927_Mono2Micro_a_practical_and_ effective_tool_for_decomposing_monolithic_Java_applications_to_microservices [2] Service Candidate Identification from Monolithic Systems Based on Execution Traces https://ieeexplore.ieee.org/document/8686152
- 10. Monolith Splitter Test Results ● Currently tested on the JPetStore [3] and AcmeAir [4] applications ● Execution traces sourced from publicly available Mono2Micro datasets on GitHub [5] ● Future testing planned on widely-used, monolithic open-source projects written in Java [3] JPetStore https://github.com/KimJongSung/jPetStore [4] AcmeAir https://github.com/acmeair/acmeair [5] Mono2Micro-FSE-2021 https://github.com/kaliaanup/Mono2Micro-FSE-2021
- 11. 5 Clusters 6 Clusters Mono2Micro JPetStore
- 12. 5 Clusters 6 Clusters FoSCI JPetStore 2 diff
- 13. 5 Clusters 6 Clusters FoSCI JPetStore 3 diff
- 14. 8 Clusters 9 Clusters Mono2Micro AcmeAir
- 15. 5 Clusters 6 Clusters FoSCI AcmeAir 2 diff
- 16. 5 Clusters 6 Clusters FoSCI AcmeAir 3 diff
- 17. Evaluation - Which approach is optimal? ● Determining the most effective clustering approach for the monolith presents a significant challenge ● Each clustering method is designed to optimize specific metrics ● The prototype includes multiple clustering approaches to enable a more informed decision when migrating a monolithic project
- 18. Current Limitations and Challenges ● Optimization of metrics computation for explainability on the way ● Varying approaches require specific input sets ● Some legacy projects lack access to Git logs ● FoSCI analysis is time-intensive on large data sets (over 10 MB) ● Capturing quality execution traces in legacy systems is challenging due to complexity
- 19. Thank you! ● Questions Reach Out for collaboration! Bruno Rossi brossi@mail.muni.cz Michal Skipala skipala3@gmail.com