Research

Bibliografia

14 min. czytania

Bibliografia

Ostatnia aktualizacja: 2026-05-28 (rozszerzono o Sekcję K: architektura serverless, +4 nowe referencje [24]-[27])

Pełna lista referencji dla projektu badawczego. Publikacje z [PDF] posiadają pełne podsumowanie w bazie wiedzy.

A. Fundamenty — Energia i Języki Programowania

[1] Pereira, R., Couto, M., Ribeiro, F., Rua, R., Cunha, J., Fernandes, J. P., & Saraiva, J. (2017). Energy efficiency across programming languages: how do energy, time, and memory relate? In Proceedings of the 10th ACM SIGPLAN International Conference on Software Language Engineering (SLE 2017). DOI: 10.1145/3136014.3136031 → publications/references/pereira-energy-efficiency-languages-2017/ Pierwsza systematyczna analiza 27 języków przez RAPL + CLBG. JavaScript (Node.js): 4.45× więcej energii niż C.

[2] Pereira, R., Couto, M., Ribeiro, F., Rua, R., Cunha, J., Fernandes, J. P., & Saraiva, J. (2021). Ranking programming languages by energy efficiency. Science of Computer Programming, vol. 205, art. 102609. Elsevier. DOI: 10.1016/j.scico.2021.102609 → publications/references/pereira-ranking-languages-energy-2021/ Rozszerzenie [1]: walidacja na Rosetta Code, analiza TOPSIS, narzędzie multi-criteria. JavaScript: 15. miejsce na 27.

[3] Cunha, S., Silva, L., Saraiva, J., & Fernandes, J. P. (2024). Trading Runtime for Energy Efficiency: Leveraging Power Caps to Save Energy across Programming Languages. In Proceedings of the 17th ACM SIGPLAN International Conference on Software Language Engineering (SLE 2024), pp. 130–142. DOI: 10.1145/3687997.3695638 → publications/references/cunha-power-caps-energy-languages-2024/ Power cap (RAPL PKGP) jako mechanizm optymalizacji — sweet spot 70-80% dla JS. Kontynuacja linii Pereira.

B. Narzędzia Pomiarowe

[4] Noureddine, A. (2022). [PDF] PowerJoular and JoularJX: Multi-Platform Software Power Monitoring Tools. 18th International Conference on Intelligent Environments (IE 2022), Biarritz, France. IEEE. DOI: 10.1109/IE54923.2022.9826760 | HAL: hal-03608223v1 → publications/with-pdf/noureddine-powerjoular-2022/ PowerJoular: CLI (Ada), per-process monitoring przez RAPL (x86) + polynomial model (ARM/RPi). Kluczowe narzędzie dla JE-3.

[17] Rehman, A., Fuerst, A., & Sharma, P. (2024). [PDF] FaasMeter: Energy-First Serverless Computing. arXiv:2408.06130 (Indiana University). DOI: 10.48550/arxiv.2408.06130 → publications/with-pdf/rehman-faasmeter-energy-2024/ Disaggregacja energii per funkcję FaaS: Kalman filter + Shapley values. Control-plane energy monitoring. Gap dla JE: zakłada runtime-agnostic energy profiles — JS runtimes (Node.js vs Bun) falsyfikują to założenie.

C. WebAssembly vs JavaScript — Energia i Wydajność

[5] De Macedo, J., Abreu, R., Pereira, R., & Saraiva, J. (2022). WebAssembly versus JavaScript: Energy and Runtime Performance. 2022 International Conference on ICT for Sustainability (ICT4S), IEEE, pp. 24–34. DOI: 10.1109/ICT4S55073.2022.00014 → publications/references/de-macedo-wasm-vs-js-energy-2022/ Pierwsza systematyczna analiza Wasm vs JS energetycznie. Wasm lepszy o ~20-30% dla CPU-bound. Tylko Node.js (V8).

D. Node.js / Deno / Bun — Porównania

[6] Smirnov, A. A., Podolskiy, E. A., Cherenkov, A. V., & Gosudarev, I. B. (2024). A comparative analysis of the performance of JavaScript code execution environments: Node.js, Deno and Bun. Программные системы и вычислительные методы (Programming Systems and Computational Methods), nr 4, pp. 109–123. DOI: 10.7256/2454-0714.2024.4.72206 → publications/references/smirnov-nodejs-deno-bun-comparison-2024/ Jedyne akademickie porównanie Node.js/Deno/Bun — tylko czas, bez energii. Gap badawczy JE-1.

[6b] Laakso, J. (2025). [PDF] The Next Generation of Server-Side JavaScript Runtimes: Node.js, Deno and Bun. Bachelor’s Thesis UAS, Turku University of Applied Sciences, Business Information Technology. 43 pp. URL: http://www.theseus.fi/bitstream/10024/905982/2/Laakso_Juuso.pdf → publications/with-pdf/laakso-js-runtimes-2025/ Grey literature (praca licencjacka). Benchmark data: Bun 103k req/s vs Node.js 73k vs Deno 73k (native); JavaScriptCore vs V8. Bez energii = gap JE-1. State of JS 2024: Node 90.8%, Bun 16.4%, Deno 11.8%.

E. WebAssembly w Edge i Serverless

[7] Marcelino, C., Copik, M., Calotoiu, A., Nastic, S., Schulz, M., & Dustdar, S. (2025). [PDF] Lumos: Performance Characterization of WebAssembly as a Serverless Runtime in the Edge-Cloud Continuum. arXiv: 2504.04570v1 → publications/with-pdf/marcelino-lumos-wasm-serverless-edge-2025/ Lumos: model wydajnościowy Wasm (Wasmtime, WasmEdge, Wasmedge AOT) na Knative. Wasm interpretowany 30-55× wolniejszy od AOT.

[8] Besozzi, M., Baresi, L., & Quattrocchi, G. (2025). [PDF] WebAssembly and Unikernels: A Comparative Study for Serverless Computing at the Edge. arXiv: 2503.08948v1 → publications/with-pdf/besozzi-wasm-unikernels-serverless-edge-2025/ Wasm cold start: 5.6ms vs Firecracker 93.5ms. Unikernels (Unikraft) jako trzecia opcja.

[9] Colosi, M., Farahani, R., Lovén, L., Prodan, R., & Villari, M. (2025). [PDF] Serverless Everywhere: A Comparative Analysis of WebAssembly Workflows Across Browser, Edge, and Cloud. arXiv: 2512.04089v1 → publications/with-pdf/colosi-serverless-everywhere-wasm-2025/ 5-krokowy DAG Rust/Wasm na browser/edge/cloud. AOT redukuje cold start o rząd wielkości. Browser lepszy dla małych payloadów.

[10] Marcelino, C., & Nastic, S. (2025). [PDF] CWASI: A WebAssembly Runtime Shim for Inter-function Communication in the Serverless Edge-Cloud Continuum. ACM/IEEE Symposium on Edge Computing (SEC ‘23), December 2023, Wilmington, DE. DOI: 10.1145/3583740.3626611 | arXiv: 2504.21503v1 → publications/with-pdf/marcelino-cwasi-wasm-shim-2025/ CWASI: trójtrybowy IPC dla co-located Wasm (embedding, UDS, Redis). 95% redukcja latencji vs WasmEdge.

F. Carbon-Aware i Sustainability-Aware Serverless

[11] Chadha, M., John, A., & Gerndt, M. (2023). [PDF] GreenCourier: Carbon-Aware Scheduling for Serverless Functions. ACM/IFIP Middleware 2023, Bologna, Italy. DOI: 10.1145/3631295.3631396 → publications/with-pdf/chadha-greencourier-carbon-serverless-2023/ Carbon-aware scheduling na OpenWhisk: carbon intensity regionów → 12% CO₂ redukcja bez degradacji latencji.

[19] Serenari, J., Sreekumar, S., Zhao, K., Sarkar, S., & Lee, S. (2024). [PDF] GreenWhisk: Emission-Aware Computing for Serverless Platform. IC2E 2024 (IEEE International Conference on Cloud Engineering). arXiv:2409.03029 DOI: 10.48550/arxiv.2409.03029 → publications/with-pdf/serenari-greenwhisk-emission-serverless-2024/ Emission-aware load balancing na OpenWhisk: real-time carbon intensity (TryCarbonara) + grid-aware function placement. Gap dla JE: zakłada stały carbon footprint per funkcja — wybór JS runtime zmienia go bezpośrednio.

[20] Qi, S., Moore, H., Hogade, N., Milojicic, D., Bash, C., & Pasricha, S. (2024). [PDF] CASA: A Framework for SLO and Carbon-Aware Autoscaling and Scheduling in Serverless Cloud Computing. IEEE IGSC 2024 (15th International Green and Sustainable Computing Conference). arXiv:2409.00550 DOI: 10.48550/arxiv.2409.00550 → publications/with-pdf/su-casa-slo-carbon-serverless-2024/ CASA: dual-objective (SLO + carbon) autoscaling z heuristic-switching (greedy + simulated annealing). 50-węzłowy klaster, Azure Function Trace. Gap dla JE: runtime-agnostic model energii — wybór JS runtime (Node.js vs Bun) zmienia carbon footprint per invocation.

[21] Sun, B., Antonopoulos, C.D., Smirni, E., Ren, B., Bellas, N., & Lalis, S. (2026). [PDF] Green or Fast? Learning to Balance Cold Starts and Idle Carbon in Serverless Computing. CCGrid 2026 (26th IEEE International Symposium on Cluster, Cloud, and Internet Computing), Sydney. arXiv:2602.23935 DOI: 10.48550/arxiv.2602.23935 → publications/with-pdf/sun-green-or-fast-serverless-2026/ LACE-RL: DQN-based adaptive keep-alive minimalizujący cold starts + idle carbon. Model: E_cold i λ_idle per function. 51.69% mniej cold starts, 77.08% mniej idle carbon vs Huawei static 60s. Gap dla JE: E_cold = P_cold × T_cold zakłada runtime-agnostic P_cold — mierzalne empirycznie dla Node.js/Deno/Bun.

[22] Tsenos, M., Peri, A., & Kalogeraki, V. (2024). [PDF] Energy Efficient Scheduling for Serverless Systems. arXiv:2410.06695 (Athens University of Economics and Business). DOI: 10.48550/arxiv.2410.06695 → publications/with-pdf/tsenos-energy-scheduling-serverless-2024/ EES: DVFS-based FaaS scheduler (PMSU + M/M/c queuing model). 4.0→3.6 GHz = 10% wolniej, 22.5% mniej energii. 28% oszczędności vs baseline. Gap dla JE: zakłada similar CPU utilization → similar energy — V8 vs JavaScriptCore falsyfikują to.

[23] Qi, S., Moore, H., Hogade, N., Milojicic, D., Bash, C., & Pasricha, S. (2024). [PDF] A Framework for SLO, Carbon, and Wastewater-Aware Sustainable FaaS Cloud Platform Management. arXiv:2410.11875 (Colorado State University + Hewlett Packard Labs). DOI: 10.48550/arxiv.2410.11875 → publications/with-pdf/qi-slo-carbon-wastewater-faas-2024/ SFCM: pierwsza triple-objective FaaS (SLO + carbon + wastewater). Local search + EA + search history. SFCM-Balance: −45%/−25%/−26% vs HYBRID. Gap dla JE: JS runtime selection jako 4. wymiar — efektywniejszy runtime redukuje energię → mniejsze cooling wastewater.

G. Cold Start i Serverless Performance

[12] Golec, M., Walia, G. K., Kumar, M., Cuadrado, F., Gill, S. S., & Uhlig, S. (2024). [PDF] Cold Start Latency in Serverless Computing: A Systematic Review, Taxonomy, and Future Directions. ACM Computing Surveys, Vol. 37, No. 4, Article 111, August 2024. DOI: 10.1145/nnnnnnn.nnnnnnn | arXiv: 2310.08437v2 → publications/with-pdf/golec-cold-start-review-2024/ Pierwsza SLR cold start. Interpretowane (JS/Node.js): krótszy cold start niż Java/.NET. Gap: brak pomiaru energii cold start.

[13] Li, J., Kulkarni, S. G., Ramakrishnan, K. K., & Li, D. (2019). [PDF] Understanding Open Source Serverless Platforms: Design Considerations and Performance. Fifth International Workshop on Serverless Computing (WOSC ‘19), Davis, CA. ACM. DOI: 10.1145/3366623.3368139 | arXiv: 1911.07449 → publications/with-pdf/li-open-source-serverless-platforms-2019/ Porównanie Knative, Kubeless, Nuclio, OpenFaaS. Throughput różni się do 10×. Kontekst dla wyboru platformy benchmarkowej.

[14] Carl, S., Schambach, M., Steinmetz, J., & Jansen, A. (2026). [PDF] Serverless Abstractions for Short-Running, Lightweight Streams. → publications/with-pdf/carl-serverless-streams-2026/ Streaming workloads w serverless — wzorce long-running vs short-running streams.

[15] Kiener, M., Chadha, M., & Gerndt, M. (2021). [PDF] Towards Demystifying Intra-Function Parallelism in Serverless Computing. Workshop on Serverless Computing (WoSC ‘21), Québec, Canada. ACM. arXiv: 2110.12090 → publications/with-pdf/kiener-intra-function-parallelism-2021/ vCPU ≠ fizyczny rdzeń. Parallelizacja oszczędza 81% kosztów (AWS Lambda). Node.js wymaga worker_threads.

H. Metodologia Eksperymentalna — Wzorce

[18] Werner, S., Borges, M. C., Wolf, K., & Tai, S. (2025). [PDF] A Comprehensive Experimentation Framework for Energy-Efficient Design of Cloud-Native Applications. 22nd IEEE International Conference on Software Architecture (ICSA’25), preprint. arXiv:2503.08641 DOI: 10.48550/arxiv.2503.08641 → publications/with-pdf/werner-experimentation-framework-energy-2025/ CLUE: framework eksperymentalny dla cloud-native na K8s z Kepler + Scaphandre + RAPL. 7-wymiarowa przestrzeń (runtime, hardware, load…). Wzorzec metodologiczny dla JE: jak kontrolować zmienne w wielowymiarowym eksperymencie energetycznym.

[16] Albonico, M., Cannizza, M. B., & Wortmann, A. (2025). Energy efficiency in ROS communication: a comparison across programming languages and workloads. Frontiers in Robotics and AI, vol. 12. DOI: 10.3389/frobt.2025.1548250 → publications/references/albonico-energy-ros-languages-2025/ C++ vs Python energia w ROS 2. Wzorzec eksperymentalny: język × częstotliwość × klienci. Python stale więcej energii.

K. Architektura i Modele Izolacji Środowisk Serverless

[24] Agache, A., Brooker, M., Iordache, A., Liguori, A., Neugebauer, R., Piwonka, P., & Popa, D.-N. (2020). Firecracker: Lightweight Virtualization for Serverless Applications. 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI 2020), pp. 419–434. URL: https://www.usenix.org/conference/nsdi20/presentation/agache → publications/references/agache-firecracker-serverless-2020/ Architektura MicroVM dla AWS Lambda/Fargate: 125ms cold start, 5MB overhead/funkcja, 150 MicroVMs/s. KVM + stripped Linux kernel. Kluczowy baseline dla analizy kosztów izolacji. Gap dla JE: brak pomiaru energii per MicroVM — koszt energetyczny izolacji nieznany.

[25] Shillaker, S., & Pietzuch, P. (2020). Faasm: Lightweight Isolation for Efficient Stateful Serverless Computing. 2020 USENIX Annual Technical Conference (ATC 2020), pp. 923–937. URL: https://www.usenix.org/conference/atc20/presentation/shillaker → publications/references/shillaker-faasm-serverless-2020/ Faasm: izolacja przez WebAssembly zamiast kontenerów dla stateful serverless. 10× szybszy cold start niż Docker. Dwa regiony pamięci: local (per-function, private) + global (shared, ograniczony). Gap dla JE: porównanie Faasm (WASM isolation) vs Node.js (process isolation) vs V8 Isolates pod kątem energii — niezbadane.

[26] Boucher, S., Kalia, A., Andersen, D. G., & Kaminsky, M. (2020). Sledge: a Serverless-first, Light-weight Wasm Runtime for the Edge. ACM/IFIP International Middleware Conference (Middleware 2020), pp. 265–279. DOI: 10.1145/3423211.3425680 → publications/references/boucher-sledge-wasm-edge-2020/ Sledge: edge serverless runtime oparty na WASM sandboxing (bez OS isolation). Cooperative scheduler + HTTP workloads na ARM. Niski overhead izolacji. Gap dla JE: model bezpośrednio porównywalny z WinterCG-compliant JS runtime — brak energii.

[27] Jangda, A., Powers, B., Berger, E. D., & Guha, A. (2019). Not So Fast: Analyzing the Performance of WebAssembly vs. Native Code. 2019 USENIX Annual Technical Conference (ATC 2019), pp. 107–120. URL: https://www.usenix.org/conference/atc19/presentation/jangda → publications/references/jangda-wasm-performance-2019/ Systematyczna analiza SPEC-CPU 2006 w WASM (Emscripten) vs native: WASM 1.55× wolniejszy średnio. Overhead pochodzi z: bezpieczeństwa pamięci (brak rejestrów segmentów), braku SIMD, braku optymalizacji link-time. Gap dla JE: brak analizy energetycznej — 1.55× wolniejszy ≠ 1.55× droższy energetycznie (DVFS, memory bandwidth).

I. Indeks cytowań i powiązań

IDAutorzyRokCytowaniaProjekt powiązany
[1]Pereira et al.2017~188JE-1, JE-3 (metodologia RAPL)
[2]Pereira et al.2021~172JE-1, JE-2 (multi-criteria ranking)
[3]Cunha et al.2024~4JE-3 (power cap)
[4]Noureddine2022~44JE-3 (narzędzie pomiarowe)
[5]De Macedo et al.2022~22JE-4 (Wasm vs JS energia)
[6]Smirnov et al.2024n/aJE-1 (gap: brak energii)
[7]Marcelino et al.2025n/aJE-4 (Wasm edge wydajność)
[8]Besozzi et al.2025n/aJE-4 (cold start Wasm)
[9]Colosi et al.2025n/aJE-3 (metodologia benchmark)
[10]Marcelino & Nastic2025n/aJE-4 (IPC Wasm, energia IPC)
[11]Chadha et al.2023n/aJE-5, JE-7 (carbon-aware)
[12]Golec et al.2024n/aJE-6 (cold start survey)
[13]Li et al.2019n/aJE-1, JE-3 (platformy serverless)
[14]Carl et al.2026n/aJE-8 (streaming workloads)
[15]Kiener et al.2021n/aJE-3 (vCPU, parallelism)
[16]Albonico et al.2025~2JE-1 (wzorzec metodologiczny)
[17]Rehman et al.2024n/aJE-10 (disaggregacja energii FaaS)
[18]Werner et al.2025n/aJE-10 (CLUE framework metodologiczny)
[19]Serenari et al.2024n/aJE-5, JE-7 (emission-aware scheduling)
[20]Qi et al. (CASA)2024n/aJE-5, JE-7 (SLO+carbon autoscaling)
[21]Sun et al.2026n/aJE-11 (LACE-RL, E_cold, λ_idle)
[22]Tsenos et al.2024n/aJE-12 (DVFS, energy×frequency)
[23]Qi et al. (SFCM)2024n/aJE-13 (triple-sustainability FaaS)
[24]Agache et al.2020~600+JE-14, JE-15 (Firecracker MicroVM baseline)
[25]Shillaker & Pietzuch2020~200+JE-14, JE-15 (WASM isolation dla serverless)
[26]Boucher et al.2020~100+JE-14, JE-15 (Sledge edge runtime)
[27]Jangda et al.2019~200+JE-14 (WASM vs native, overhead izolacji)

J. Format cytowań (ACM/IEEE)

Dla pracy doktorskiej preferowany format IEEE:

[1] R. Pereira, M. Couto, F. Ribeiro, R. Rua, J. Cunha, J. P. Fernandes, and J. Saraiva,
    "Energy efficiency across programming languages: how do energy, time, and memory relate?"
    in Proc. 10th ACM SIGPLAN Int. Conf. Software Language Engineering (SLE), 2017.
    DOI: 10.1145/3136014.3136031

[2] R. Pereira et al., "Ranking programming languages by energy efficiency,"
    Science of Computer Programming, vol. 205, p. 102609, 2021.
    DOI: 10.1016/j.scico.2021.102609

[3] S. Cunha, L. Silva, J. Saraiva, and J. P. Fernandes,
    "Trading Runtime for Energy Efficiency: Leveraging Power Caps to Save Energy across Programming Languages,"
    in Proc. 17th ACM SIGPLAN Int. Conf. Software Language Engineering (SLE), 2024, pp. 130–142.
    DOI: 10.1145/3687997.3695638

[4] A. Noureddine, "PowerJoular and JoularJX: Multi-Platform Software Power Monitoring Tools,"
    in Proc. 18th Int. Conf. Intelligent Environments (IE 2022), 2022.
    DOI: 10.1109/IE54923.2022.9826760

[5] J. De Macedo, R. Abreu, R. Pereira, and J. Saraiva,
    "WebAssembly versus JavaScript: Energy and Runtime Performance,"
    in Proc. Int. Conf. ICT for Sustainability (ICT4S), 2022, pp. 24–34.
    DOI: 10.1109/ICT4S55073.2022.00014

[6] A. A. Smirnov, E. A. Podolskiy, A. V. Cherenkov, and I. B. Gosudarev,
    "A comparative analysis of the performance of JavaScript code execution environments: Node.js, Deno and Bun,"
    Programming Systems and Computational Methods, no. 4, pp. 109–123, 2024.
    DOI: 10.7256/2454-0714.2024.4.72206

[7] C. Marcelino et al., "Lumos: Performance Characterization of WebAssembly as a Serverless Runtime
    in the Edge-Cloud Continuum," arXiv:2504.04570, 2025.

[8] M. Besozzi, L. Baresi, and G. Quattrocchi,
    "WebAssembly and Unikernels: A Comparative Study for Serverless Computing at the Edge,"
    arXiv:2503.08948, 2025.

[9] M. Colosi, R. Farahani, L. Lovén, R. Prodan, and M. Villari,
    "Serverless Everywhere: A Comparative Analysis of WebAssembly Workflows Across Browser, Edge, and Cloud,"
    arXiv:2512.04089, 2025.

[10] C. Marcelino and S. Nastic, "CWASI: A WebAssembly Runtime Shim for Inter-function Communication
     in the Serverless Edge-Cloud Continuum,"
     in Proc. 8th ACM/IEEE Symp. Edge Computing (SEC '23), 2023.
     DOI: 10.1145/3583740.3626611

[11] M. Chadha, A. John, and M. Gerndt, "GreenCourier: Carbon-Aware Scheduling for Serverless Functions,"
     in Proc. ACM/IFIP Middleware, 2023.
     DOI: 10.1145/3631295.3631396

[12] M. Golec et al., "Cold Start Latency in Serverless Computing: A Systematic Review, Taxonomy,
     and Future Directions," ACM Computing Surveys, vol. 37, no. 4, art. 111, Aug. 2024.

[13] J. Li, S. G. Kulkarni, K. K. Ramakrishnan, and D. Li,
     "Understanding Open Source Serverless Platforms: Design Considerations and Performance,"
     in Proc. 5th Int. Workshop Serverless Computing (WOSC '19), 2019.
     DOI: 10.1145/3366623.3368139

[14] S. Carl, M. Schambach, J. Steinmetz, and A. Jansen,
     "Serverless Abstractions for Short-Running, Lightweight Streams," 2026.

[15] M. Kiener, M. Chadha, and M. Gerndt,
     "Towards Demystifying Intra-Function Parallelism in Serverless Computing,"
     in Proc. Workshop Serverless Computing (WoSC '21), 2021.
     arXiv:2110.12090

[16] M. Albonico, M. B. Cannizza, and A. Wortmann,
     "Energy efficiency in ROS communication: a comparison across programming languages and workloads,"
     Frontiers in Robotics and AI, vol. 12, 2025.
     DOI: 10.3389/frobt.2025.1548250

[17] A. Rehman, A. Fuerst, and P. Sharma,
     "FaasMeter: Energy-First Serverless Computing,"
     arXiv:2408.06130, 2024.

[18] S. Werner, M. C. Borges, K. Wolf, and S. Tai,
     "A Comprehensive Experimentation Framework for Energy-Efficient Design of Cloud-Native Applications,"
     in Proc. 22nd IEEE Int. Conf. Software Architecture (ICSA), 2025.
     arXiv:2503.08641

[19] J. Serenari, S. Sreekumar, K. Zhao, S. Sarkar, and S. Lee,
     "GreenWhisk: Emission-Aware Computing for Serverless Platform,"
     in Proc. IEEE Int. Conf. Cloud Engineering (IC2E), 2024.
     arXiv:2409.03029

[20] S. Qi, H. Moore, N. Hogade, D. Milojicic, C. Bash, and S. Pasricha,
     "CASA: A Framework for SLO and Carbon-Aware Autoscaling and Scheduling in Serverless Cloud Computing,"
     in Proc. 15th IEEE Int. Green and Sustainable Computing Conf. (IGSC), 2024.
     arXiv:2409.00550

[21] B. Sun, C. D. Antonopoulos, E. Smirni, B. Ren, N. Bellas, and S. Lalis,
     "Green or Fast? Learning to Balance Cold Starts and Idle Carbon in Serverless Computing,"
     in Proc. 26th IEEE Int. Symp. Cluster, Cloud, and Internet Computing (CCGrid), 2026.
     arXiv:2602.23935

[22] M. Tsenos, A. Peri, and V. Kalogeraki,
     "Energy Efficient Scheduling for Serverless Systems,"
     arXiv:2410.06695, 2024.

[23] S. Qi, H. Moore, N. Hogade, D. Milojicic, C. Bash, and S. Pasricha,
     "A Framework for SLO, Carbon, and Wastewater-Aware Sustainable FaaS Cloud Platform Management,"
     arXiv:2410.11875, 2024.

[24] A. Agache, M. Brooker, A. Iordache, A. Liguori, R. Neugebauer, P. Piwonka, and D.-N. Popa,
     "Firecracker: Lightweight Virtualization for Serverless Applications,"
     in Proc. 17th USENIX Symp. Networked Systems Design and Implementation (NSDI), 2020, pp. 419–434.

[25] S. Shillaker and P. Pietzuch,
     "Faasm: Lightweight Isolation for Efficient Stateful Serverless Computing,"
     in Proc. 2020 USENIX Annual Technical Conference (ATC), 2020, pp. 923–937.

[26] S. Boucher, A. Kalia, D. G. Andersen, and M. Kaminsky,
     "Sledge: a Serverless-first, Light-weight Wasm Runtime for the Edge,"
     in Proc. ACM/IFIP Int. Middleware Conf. (Middleware), 2020, pp. 265–279.
     DOI: 10.1145/3423211.3425680

[27] A. Jangda, B. Powers, E. D. Berger, and A. Guha,
     "Not So Fast: Analyzing the Performance of WebAssembly vs. Native Code,"
     in Proc. 2019 USENIX Annual Technical Conference (ATC), 2019, pp. 107–120.

Graf