In a paper published in Cluster Computing, Jasmine Kaur, Inderveer Chana and Anju Bala introduce an enhanced version of a serverless simulation framework that adds a virtual machine layer between physical servers and the containers that actually run user code. On top of that more realistic architecture, they bolt on a multi-agent reinforcement learning technique known as proximal policy optimization, or PPO, to decide where workloads should go. The results are striking: cold start latency drops by up to 29 percent, the number of active physical machines falls by as much as 66 percent, and energy consumption shrinks by up to 23 percent. For an industry under mounting pressure to curb the enormous electricity appetite of data centers, those numbers carry real weight.

To understand why the virtual machine layer matters, it helps to look at how serverless platforms are actually built. In principle, containers, lightweight packages of software that share the host operating system’s kernel, could run directly on physical machines. That is precisely how ServlessSimPro, the earlier simulation platform on which the new work builds, modeled the world. But in production deployments at major cloud providers and in popular open-source platforms such as Knative and OpenFaaS, containers are almost always nested inside virtual machines. The VM adds an extra boundary of isolation and security, but it also introduces an additional layer of resource management, with its own scheduling decisions, consolidation opportunities and performance overheads.

By ignoring that layer, earlier simulators produced results that diverged from what operators actually observe. A scheduling algorithm that looks brilliant in a flat, container-on-metal simulation may behave very differently when it must first choose a virtual machine, and only then a container within it. The enhanced framework bridges this architectural gap, giving researchers a testbed whose resource abstractions mirror the deployment models used in the real world. That matters because simulation is where most scheduling research lives: few academic groups can run experiments on hyperscale infrastructure, so the fidelity of the simulator effectively determines how transferable their ideas are.

The second half of the contribution is the scheduling brain itself. Rather than a single centralized controller making every placement decision, the researchers deploy multiple reinforcement learning agents, each powered by proximal policy optimization, working in parallel across the simulated cluster. PPO is a policy-gradient method that has become a favorite of the reinforcement learning community because it makes steady, clipped updates to its decision policy, avoiding the catastrophic performance collapses that can plague more aggressive learning algorithms. Each agent is responsible for a slice of the scheduling problem: deciding where containers should be placed, when virtual machines should be consolidated onto fewer physical hosts, and how incoming workloads should be distributed across the fleet.

Decentralization is not just an aesthetic choice. In a large data center, a single scheduler becomes a bottleneck and a single point of failure, and the state of the system grows so large that no one agent can see everything at once. Multi-agent architectures let decisions happen closer to the resources they affect, and the multi-agent formulation of PPO, often abbreviated MAPPO, allows the agents to learn coordinated behavior despite their partial views. The approach builds on the authors’ earlier work on multi-agent deep Q-learning for serverless job scheduling, but PPO’s more stable training dynamics make it better suited to the high-dimensional, continuous decisions involved in juggling containers, VMs and workloads simultaneously.

The headline metric, cold start latency, deserves particular attention because it is the Achilles’ heel of serverless computing. When a function is invoked and no warm instance exists, the platform must provision a container from scratch, an operation that can add hundreds of milliseconds or more to response time. For latency-sensitive applications, from interactive web services to real-time data pipelines, those milliseconds are the difference between a seamless experience and a visibly sluggish one. By learning placement and consolidation policies that keep likely-to-be-needed functions warm and co-located with available capacity, the MAPPO-driven scheduler cuts cold starts by nearly a third in the reported experiments.

The efficiency gains are equally significant. Consolidating workloads so that up to 66 percent fewer physical machines need to stay active is a dramatic reduction, and it cascades directly into the 23 percent cut in energy consumption. Data centers are among the fastest-growing consumers of electricity worldwide, and idle servers draw substantial power even when they process nothing. Any technique that lets operators switch off more of the fleet, without degrading the responsiveness users feel, translates into both lower costs and lower carbon emissions. The fact that the framework achieves these savings while simultaneously improving latency suggests the agents are finding genuinely smarter trade-offs, not merely shifting costs from one column to another.

The study also situates itself within a rapidly maturing ecosystem of simulation tools. Open-source options such as FaasSim, FaaS-Sim and SimFaaS have given researchers starting points for modeling function-as-a-service platforms, and commercial offerings from AWS, Microsoft Azure, Google Cloud and IBM have defined the de facto behavior the simulators try to capture. What has been scarce, the authors argue, is a comprehensive platform that combines realistic multi-layer resource modeling with the ability to train and evaluate learning-based schedulers. By pairing the VM-aware architecture with MAPPO, the framework becomes both a measurement instrument and a training ground, letting researchers prototype scheduling policies that can be benchmarked consistently before any real-world trial.

The broader significance of the work lies in the convergence of two trends: the industrial consolidation around serverless architectures and the rapid advance of multi-agent reinforcement learning as a practical tool for systems management. As edge computing, Kubernetes-based orchestration and energy-aware provisioning continue to reshape how cloud infrastructure is organized, the ability to simulate those systems faithfully, and to let learning agents discover scheduling policies that human engineers might never hand-design, could define the next generation of cloud efficiency. The Thapar Institute team’s framework offers the research community a way to explore that future at scale, on a laptop rather than a data center, and with an architectural honesty that earlier tools lacked. If the reported gains hold when such policies migrate from simulation to production clusters, the invisible machinery of the cloud may soon be run by algorithms that learned their craft in a simulator, one carefully modeled virtual machine at a time.

Subject of Research: VM-aware serverless simulation and multi-agent reinforcement learning for cloud job scheduling

Article Title: VM-aware serverless simulation framework with multi-agent proximal policy optimization for scalable and efficient job scheduling

Article References: Kaur, J., Chana, I., & Bala, A. (2026). VM-aware serverless simulation framework with multi-agent proximal policy optimization for scalable and efficient job scheduling. Cluster Computing, 29(13), Article 758. https://doi.org/10.1007/s10586-026-06585-w

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06585-w

Keywords: serverless computing, simulation framework, virtual machines, multi-agent reinforcement learning, proximal policy optimization, job scheduling, cold start latency, energy efficiency, cloud computing, resource utilization, container orchestration, data centers

The conventional method for assessing mycorrhizal colonization, known as the gridline intersect method, has remained the standard for decades. It requires fixing, dyeing and microscopically examining host plant roots to determine the percentage of root length containing fungal structures such as mycelia, vesicles and arbuscules. The approach is not only time-consuming, laborious and costly, but also prone to error, since staining can color root cortex vessels in ways that mimic false mycorrhizal structures. More fundamentally, the method cannot distinguish between natural, neutral and even parasitic symbiotic relationships, and it ignores entirely the question of fungal effectiveness, that is, whether the colonization actually improves the plant’s performance. Earlier attempts to find alternatives, including the magnified intersections method and the detection of mycorrhiza-specific isozymes, improved objectivity but still fell short of quantifying the functional benefits of the partnership.

The new approach flips the problem on its head. Rather than counting fungal organs inside roots, Jahan and Nassiri-Mahallati measured the eco-physiological consequences of symbiosis in the plant itself. Their reasoning was straightforward: if the fungus genuinely benefits its host, that benefit should reveal itself in improved growth, better resource capture and higher yield. By measuring a comprehensive suite of plant and soil traits in inoculated and non-inoculated maize, and then applying mathematical modeling to the resulting data, the researchers could trace exactly which plant functions the fungi enhanced and by how much. This multidisciplinary framework draws together plant ecophysiology, soil ecology, multivariate statistics and mathematical modeling into a single analytical pipeline that goes beyond observation to causal understanding.

The evidence base came from two years of field experiments conducted at the research farm of Ferdowsi University of Mashhad, in the semi-arid Kashaf River watershed of northeastern Iran, where mean annual precipitation is just 252 millimeters. The experiments followed a split-plot arrangement within a randomized complete block design with three replications, spanning four different cropping systems ranging from high-input to ecological management. Maize seeds of the single cross 704 cultivar were coated with powdered inoculum containing propagules of two fungal species, Rhizophagus intraradices and Funneliformis mosseae, at a concentration of one million colony-forming units per gram, while control seeds were left uninoculated. The team then measured everything from leaf area index, dry matter yield and seed yield to maximum photosynthesis rate, leaf chlorophyll content via SPAD readings, canopy temperature, plant phosphorus content, specific root length, soil respiration and the chlorophyll fluorescence ratio Fv/Fm.

Analysis of variance confirmed that mycorrhizal inoculation significantly affected nearly every trait measured, with mycorrhizal plants showing significantly higher values than their non-mycorrhizal counterparts for maximum photosynthesis rate, leaf area index, dry matter, seed yield, plant height, stem diameter, soil respiration, specific root length, root colonization, SPAD readings and plant phosphorus. But the real innovation lay in what came next. Confirmatory factor analysis divided the fourteen measured variables into two coherent groups, which the researchers named on eco-physiological grounds. The first factor, containing leaf area index, SPAD readings, dry matter, root colonization and stem diameter, was labeled the resource capture construct. The second, comprising specific root length, plant height, maximum photosynthesis rate, canopy temperature, cob number, seed yield, plant phosphorus, soil respiration and Fv/Fm, was named the resource utilization construct. The reliability of both constructs was confirmed by Cronbach’s Alpha values of 0.77 and 0.74 respectively, exceeding the conventional threshold of 0.70.

With the two latent constructs established, the team employed structural equation modeling, a statistical technique that tests networks of hypothesized causal relationships against observed covariance patterns. SEM has grown increasingly popular in ecology since 2000 precisely because it can partition direct and indirect effects along multiple pathways, something regression and analysis of covariance cannot achieve when the goal is understanding biological mechanisms. The fitted model proposed a causal path running from resource capture to resource utilization, with the capture construct driving the utilization construct and both jointly shaping final yield. The model’s goodness of fit was assessed through several indices: a normalized chi-square of 2.33, comfortably within the accepted range of one to three, and a coefficient of determination of R squared equal to 0.37, which by the classification standards of Hair and colleagues falls within the good range of 0.25 to 0.50.

That R squared value carries the study’s headline finding: thirty-seven percent of the total variance in maize performance can be explained by resource capture and utilization through the mycorrhizal collaboration. In other words, the direct advantages of AMF symbiosis, quantified through the eco-physiological machinery of the plant, account for a substantial and now measurable share of what makes an inoculated crop thrive. The standardized path coefficients revealed which traits mattered most. Leaf area index showed a direct effect of 0.888 on the resource capture construct, meaning that a one standard deviation increase in capture translates into a 0.888 standard deviation increase in leaf area. Dry matter showed a squared multiple correlation of 0.718 with resource utilization, indicating that the construct explains nearly seventy-two percent of dry matter variation. Stem diameter ranked highest after leaf area and dry matter at 0.707, reflecting its role in the vascular transport of assimilates between sources and sinks.

The placement of root colonization and SPAD readings within the resource capture construct proved particularly telling. From an eco-physiological standpoint, mycorrhizal colonization of roots increased leaf chlorophyll content, which in turn plays an essential role in capturing and utilizing radiation in the photosynthetic system. Plant height correlated strongly with specific root length, canopy temperature and leaf area index, defining the canopy architecture that governs how efficiently a crop intercepts sunlight. The analysis also showed that the effect of AMF on yield was largely indirect, mediated through increases in leaf area and plant height rather than acting directly on yield itself. Meanwhile, the negative effect of resource utilization on specific root length, a coefficient of minus 0.534, is consistent with a well-known trade-off: when fungal hyphae effectively extend the root’s absorption network, the plant can afford to invest less in its own root length per unit of soil volume.

The practical implications extend well beyond maize. Because the method relies on farm-measured traits rather than laboratory microscopy, it can in principle be applied to most crop plants, with results varying by plant species, fungal species and environmental conditions. The authors suggest that identifying the plant characteristics involved in symbiosis makes it possible to manage and strengthen them, improving final performance. As an agroecological management tool, for example, farmers could synchronize the timing of maximum leaf area index with the long sunny days of summer by adjusting planting density, planting date, irrigation and fertilization, thereby maximizing the efficiency of the photosynthetic system when radiation capture is at its peak. The approach could also help quantify the effects of inputs and treatments on yield formation across different cropping systems, since structural equation modeling works on correlation and variance-covariance matrices that can pool data across management regimes.

The study does carry caveats. The experiments evaluated the effect of two exogenous fungal species on the plant, not the entire indigenous AMF community in the soil, and native fungi could not be eliminated from the field, so the reported functions rest on exogenous measurements alone. The RMSEA value of 0.169, while above the commonly cited cut-off of 0.06, is considered mediocre in models with small samples and few variables, where the index tends to inflate; the authors note that chi-square provides a sufficient measure of model accuracy in such cases. Still, the broader vision is ambitious: a fast, reliable and low-cost method to identify causal paths and quantify symbiotic efficacy, giving farmers, advisors, researchers and policymakers a practical foundation for deploying mycorrhizal technology across agroecosystems. At a time when phosphorus fertilizers face supply restrictions, rapid fixation in soils, high costs and pollution concerns, a tool that measures exactly what fungi contribute to crop productivity, while pointing the way to reduced inputs, cost and labor, may prove one of the more quietly revolutionary developments in sustainable agriculture.

Subject of Research: Quantifying the eco-physiological benefits of arbuscular mycorrhizal fungal symbiosis with maize using structural equation modeling

Article Title: Discovering and quantifying the eco-physiological advantages of plant-soil-Arbuscular Mycorrhizal Fungi (AMF) system: a promising eco-math-statistical modelling approach

Article References: Jahan, M., & Nassiri-Mahallati, M. (2025). Discovering and quantifying the eco-physiological advantages of plant-soil-Arbuscular Mycorrhizal Fungi (AMF) system: a promising eco-math-statistical modelling approach. Discover Ecology, 1(1), Article 2. https://doi.org/10.1007/s44396-025-00001-0

Image Credits: AI Generated

DOI: 10.1007/s44396-025-00001-0

Keywords: arbuscular mycorrhizal fungi, maize, symbiosis, structural equation modeling, eco-physiology, resource capture, resource utilization, sustainable agriculture, soil ecology, phosphorus uptake, crop yield, agroecology