Cloud computing research has a fundamental problem: testing resource management algorithms, scheduling policies, and provisioning strategies on real cloud infrastructure is expensive, difficult to reproduce, and impossible to scale freely. A single experiment on Amazon EC2 or Microsoft Azure costs real money, takes unpredictable time, and cannot be precisely repeated — two critical requirements in scientific research.
CloudSim solves this problem. As the most widely cited and most widely used open-source cloud computing simulation toolkit in the world, CloudSim allows researchers, developers, and engineers to model and simulate entire cloud environments — data centers, virtual machines, workloads, network topologies, and resource policies — on a single laptop, for free, with full reproducibility and control.
This in-depth review covers everything you need to know about CloudSim in 2026: what it is, how it works, what it does well, where it falls short, how it compares to alternatives, and who should — and shouldn’t — use it.
CloudSim is a free, open-source, Java-based simulation framework for modeling and simulating cloud computing environments, infrastructure, and services. It was developed primarily at the Cloud Computing and Distributed Systems (CLOUDS) Laboratory at the University of Melbourne, Australia, under the leadership of Professor Rajkumar Buyya — one of the world’s most cited cloud computing researchers.
In simple terms, CloudSim lets you simulate a complete cloud computing infrastructure — including data centers, servers, virtual machines, and application workloads — entirely in software, without spending a single dollar on real cloud resources. You can test how different scheduling algorithms, resource allocation policies, and virtual machine placement strategies perform under varying workloads and configurations, and get precise, reproducible results you can publish, cite, and build upon.
CloudSim has become one of the most referenced tools in cloud computing research globally, cited in thousands of peer-reviewed papers across IEEE, ACM, Springer, Elsevier, and Wiley publications. Organizations including HP Labs have used it in investigations of cloud resource provisioning and energy-efficient data center management.
The latest stable version is CloudSim v6.0.0-beta, with CloudSim 7G — a next-generation toolkit for future cloud environments — published in a 2025 paper in the Software: Practice and Experience journal.
In one sentence: CloudSim is the standard simulation toolkit for cloud computing research — trusted by universities, research labs, and developers worldwide for testing new cloud algorithms before touching real infrastructure.
Understanding CloudSim’s evolution helps explain both its strengths and its limitations.
| Year | Milestone |
|---|---|
| 2008 | CloudSim development begins at CLOUDS Lab, University of Melbourne |
| 2009 | First public presentation at HPCS 2009, Leipzig, Germany |
| 2011 | Landmark paper published in Software: Practice and Experience (Wiley), establishing CloudSim as the standard citation in cloud simulation research |
| 2012–2015 | Multiple extensions created by independent researchers — NetworkCloudSim, CloudAnalyst, WorkflowSim, iFogSim |
| 2017 | CloudSim Plus published — a re-engineered fork addressing software engineering weaknesses in the original codebase |
| 2019–2022 | Continued extension development; CloudSim 4.x and 5.x releases; growing ecosystem of specialized variants |
| 2025 | CloudSim 7G paper published — presenting an integrated toolkit for modeling future generation cloud environments |
| 2026 | Current stable version: CloudSim v6.0.0-beta; CloudSim Plus continues as the state-of-the-art maintained fork |
CloudSim’s architecture was deliberately inspired by the limitations of real cloud testbeds — platforms like Amazon EC2 that limit experiments to testbed scale and make reproducibility difficult. The goal from the start was a generalized, extensible framework where researchers could focus on the specific system design question they wanted to investigate without worrying about low-level infrastructure management.
CloudSim is built on a discrete event simulation (DES) engine, which means it models the passage of time as a sequence of discrete events rather than as a continuous flow. Each event — a VM being allocated, a cloudlet starting execution, a host going offline — is processed in sequence, building a complete picture of how the simulated system would behave over time.
CloudSim’s architecture is organized into four conceptual layers:
Layer 1 — Simulation Core (SimJava / CloudSim Engine) The foundational layer handles event management, simulation clock, and inter-entity messaging. All simulation entities — DataCenters, Brokers, VMs — communicate through this event-driven messaging system.
Layer 2 — Cloud Layer Contains the primary modeling classes: DataCenter, Host, VM, Cloudlet, DatacenterBroker, and CloudInformationService. This is where the core simulation logic lives — resource allocation, scheduling policies, and VM lifecycle management.
Layer 3 — User Code Layer Where researchers and developers write their simulation scenarios — defining infrastructure configurations, workloads, and the policies they want to test.
Layer 4 — Extension Layer Where custom policies, new resource models, and domain-specific extensions (like iFogSim for fog computing or GPUCloudSim for GPU workloads) plug into the framework.
A typical CloudSim simulation follows this event sequence:
This clean, predictable flow is one of CloudSim’s most praised characteristics — it makes the simulation logic transparent and the results interpretable.
CloudSim’s feature set has expanded significantly since its initial release. The following are the core capabilities supported by the current version:
CloudSim supports the modeling and simulation of large-scale cloud data center infrastructure — including thousands of physical hosts, storage resources, and network connections — on a single physical machine. This is one of its most important capabilities: you can simulate the behavior of a data center at a scale that would be prohibitively expensive to replicate in a real environment.
CloudSim provides a complete virtualization engine for creating, deploying, migrating, and destroying virtual machines within simulated data centers. Policies for provisioning host resources to VMs are fully customizable — you can implement your own allocation strategy and test it against any workload.
Two fundamental scheduling models are supported out of the box:
Researchers can implement and plug in custom scheduling algorithms — round-robin, priority-based, machine learning-driven schedulers — and compare their performance against baseline policies within controlled simulation conditions.
CloudSim includes models for energy-aware computational resources, enabling simulation of power consumption in data centers. Researchers can evaluate how different VM placement strategies, consolidation policies, or workload scheduling decisions affect energy usage — a critical dimension for green cloud computing research.
Network topology modeling allows simulation of message-passing applications and bandwidth-constrained workloads across geographically distributed data center environments. Researchers studying latency-sensitive applications, distributed storage systems, or cloud-edge interactions can model the network layer alongside the compute layer.
CloudSim supports both single cloud and inter-networked cloud (federation) environments, allowing simulation of scenarios where multiple cloud providers collaborate to meet user demand. The federated cloud model has been shown in CloudSim case studies to significantly improve application QoS under fluctuating resource and service demand patterns.
Alongside virtual machines, CloudSim supports modeling and simulation of application containers — reflecting the industry shift toward containerized workloads (Docker, Kubernetes) that has accelerated in cloud deployments throughout the 2020s.
CloudSim supports dynamic insertion of simulation entities, as well as stop and resume functionality during a running simulation — allowing researchers to model dynamic cloud environments where infrastructure and workloads change in real time.
The framework includes support for SLA-aware provisioning, enabling simulation of cloud scenarios where providers must meet committed performance guarantees and where SLA violations have associated cost penalties.
One of CloudSim’s most frequently cited advantages is how accessible it is to set up. You do not need high-specification hardware, a paid license, or a complex installation process.
| Requirement | Detail |
|---|---|
| Operating System | Windows, macOS, or Linux |
| Java Development Kit (JDK) | JDK 8 or higher (JDK 11+ recommended) |
| IDE | Eclipse, IntelliJ IDEA, or NetBeans (Eclipse most commonly used in tutorials) |
| Maven or Gradle | For dependency management (optional but recommended) |
| Hardware | Any standard laptop or desktop — no high-spec requirements |
Step 1: Install Java Development Kit (JDK) Download and install JDK 11 or higher from the Oracle website or use an OpenJDK distribution. Set the JAVA_HOME environment variable to point to your JDK installation.
Step 2: Download CloudSim Clone the CloudSim repository from GitHub at https://github.com/Cloudslab/cloudsim or download the latest release package (JAR files, source code, and examples). The package includes the core CloudSim JAR, examples demonstrating common simulation patterns, and full API documentation (Javadoc).
Step 3: Set Up Your IDE Open Eclipse or IntelliJ IDEA and create a new Java project. Add the CloudSim JAR to your project’s build path (in Eclipse: right-click project > Build Path > Add External JARs). If using Maven, add the CloudSim dependency to your pom.xml.
Step 4: Run the Example Simulations CloudSim ships with a set of example simulation files (e.g., CloudSimExample1.java through CloudSimExample8.java) that demonstrate increasingly complex scenarios — from a single VM and cloudlet to multi-datacenter federated environments. Running these examples first is the standard recommended onboarding path.
Step 5: Build Your Custom Simulation Once you are comfortable with the example structure, begin building your own simulation scenario — defining your data center configuration, VM profiles, workloads, and the specific resource management policy you want to test.
Every CloudSim simulation follows a consistent programmatic workflow:
int num_user = 1;
Calendar calendar = Calendar.getInstance();
boolean trace_flag = false;
CloudSim.init(num_user, calendar, trace_flag);
Initialize the simulation environment by setting the number of simulated users, the simulation start time, and whether to enable detailed event tracing.
Define your cloud infrastructure — the number of hosts, processing elements per host, CPU speed (MIPS), RAM, storage, and bandwidth. Specify the VM allocation policy (e.g., VmAllocationPolicySimple).
Datacenter datacenter0 = createDatacenter("Datacenter_0");
The broker acts as the intermediary between users and data centers — responsible for VM allocation and cloudlet scheduling.
DatacenterBroker broker = new DatacenterBroker("Broker");
int brokerId = broker.getId();
Define VM configurations — number of CPUs, MIPS, RAM, storage, bandwidth, OS, and VMM — and submit the list to the broker.
Vm vm = new Vm(vmid, brokerId, mips, pesNumber, ram, bw, size, vmm, new CloudletSchedulerTimeShared());
broker.submitVmList(vmlist);
Cloudlets represent the computational tasks (jobs) to be executed on the VMs. Define their length, required PEs, input/output file sizes, and utilization models.
Cloudlet cloudlet = new Cloudlet(id, length, pesNumber, fileSize, outputSize, utilizationModel, utilizationModel, utilizationModel);
broker.submitCloudletList(cloudletList);
CloudSim.startSimulation();
List<Cloudlet> newList = broker.getCloudletReceivedList();
CloudSim.stopSimulation();
printCloudletList(newList);
The output shows each cloudlet’s execution details — the VM it ran on, start time, finish time, and execution duration — forming the raw data for your analysis.
Understanding CloudSim’s core class structure is essential for productive use of the toolkit.
The master simulation class — initializes and starts the simulation environment, manages the simulation clock, and stops execution once all entities have been processed.
A registry service that maintains a list of all active cloud resources (data centers) in the simulation. DatacenterBrokers query the CIS to discover available infrastructure.
Models a physical cloud data center — its hosts, storage, bandwidth, cost model, and associated resource allocation policies. Multiple DataCenters can be created to simulate federated or geographically distributed cloud environments.
Represents a physical server within a DataCenter. Configurable with the number of processing elements (PEs), MIPS rating per PE, RAM, storage, and bandwidth. Hosts apply VM scheduling policies (time-shared or space-shared) to allocate resources to running VMs.
Models a virtual machine instance with configurable vCPUs, MIPS, RAM, storage, bandwidth, and a cloudlet scheduler determining how tasks are executed on the VM.
The fundamental unit of work in CloudSim — representing a computational task submitted by a user. Each cloudlet has a defined length (in MI — Million Instructions), required PEs, file sizes, and utilization models describing how it consumes CPU, RAM, and bandwidth over time.
The intelligent intermediary that maps VMs to available DataCenters and maps Cloudlets to available VMs according to its scheduling strategy. Researchers typically extend this class to implement custom task scheduling and resource allocation algorithms.
Defines how a cloudlet utilizes resources over time — full utilization (UtilizationModelFull), stochastic (UtilizationModelStochastic), or custom patterns. This is a key customization point for modeling realistic, variable workloads.
Controls how VMs are placed on physical hosts within a DataCenter — the policy determines which host receives each incoming VM. Implementing and testing a novel VM placement algorithm typically involves extending this class.
Controls how host processing resources are shared among VMs running on a single host — time-shared or space-shared. Determines whether VMs compete for CPU resources or receive exclusive allocations.
One of CloudSim’s most significant assets is its extensibility. An ecosystem of community-developed extensions has expanded its applicability well beyond the original IaaS simulation scope.
A visual, GUI-based tool built on CloudSim that allows modelling and analysis of cloud computing environments without writing Java code. Particularly useful for educators and researchers who want to explore simulation scenarios interactively before committing to programmatic implementation.
Extends CloudSim with detailed network simulation capabilities for modeling parallel, message-passing applications across distributed cloud infrastructure.
An extension for simulating Fog and Edge Computing environments — modeling resource management across edge devices, cloudlets (small local data centers), and cloud back-ends. Essential for researchers working on IoT, latency-sensitive edge applications, and mobile edge computing.
Focused specifically on edge computing scenarios, EdgeCloudSim adds mobile device mobility modeling, WLAN and WAN network modeling, and realistic load generation for evaluating Mobile Edge Computing (MEC) policies.
An extension for simulating GPU-accelerated cloud infrastructures — supporting multi-GPU data center setups, customizable GPU scheduling policies, and GPU remoting simulation.
Adds support for scientific workflow simulation — modeling complex, dependency-ordered computational workflows typical in e-science, bioinformatics, and high-performance computing research.
Introduces performance heterogeneity and uncertainty into CloudSim simulations — modeling variable VM performance, straggler machines, and task failures. Essential for research on fault tolerance and performance variability in shared cloud infrastructure.
A command-line tool allowing CloudSim simulation scenarios to be defined in YAML configuration files rather than Java code — making it easier for non-programmers to configure and run simulations.
Extends CloudSim with thermal modeling — incorporating physical temperature characteristics of data center nodes and using deep learning-based temperature prediction for cloud resource management research.
Enables distributed multi-server simulation by leveraging the Hazelcast framework — allowing very large-scale simulations to be distributed across multiple physical machines.
Extends CloudSim Plus to model synchronous hierarchical architectures such as Elasticsearch — used in research on large-scale distributed data systems.
CloudSim Plus deserves its own section because many researchers in 2026 use it instead of — or as a direct upgrade to — the original CloudSim.
CloudSim Plus is a completely re-engineered fork of CloudSim 3, rebuilt to address significant software engineering weaknesses in the original codebase. It is currently maintained as a Java 25-compatible framework, making it the state-of-the-art cloud simulation toolkit as of 2026.
The original CloudSim codebase accumulated technical debt over years of development — code duplication, poor separation of concerns, low testability, and limited extensibility. CloudSim Plus was created to address these issues systematically, applying modern software engineering principles including SOLID design, design patterns, comprehensive unit testing, and continuous integration.
| Feature | CloudSim (Original) | CloudSim Plus |
|---|---|---|
| Java Compatibility | Java 8 | Java 25 |
| Active Maintenance | Limited (v6.0.0-beta) | Actively maintained |
| Code Quality | Legacy issues (duplication, coupling) | Refactored for SOLID principles |
| Documentation | Moderate | Extensive (Javadoc + wiki) |
| Unit Tests | Minimal | Comprehensive test suite |
| Exclusive Features | Basic | Multi-cloud, vertical/horizontal VM scaling, fault injection |
| Performance | Baseline | Faster simulation execution |
| Community Support | Declining | Active (GitHub Discussions + Google Group) |
| Ease of Extension | Moderate difficulty | Significantly easier |
| Python Support | No | No |
Recommendation for new users: If you are starting a new cloud simulation project in 2026, CloudSim Plus is the recommended choice over the original CloudSim. It is more accurate, actively maintained, better documented, and significantly easier to extend.
CloudSim has been applied across a remarkably broad range of cloud computing research domains since its introduction.
The most common application — testing novel VM scheduling algorithms, cloudlet scheduling policies, resource allocation strategies, and auto-scaling mechanisms in controlled, repeatable conditions. Researchers implement their proposed algorithm as a custom DatacenterBroker or VmAllocationPolicy subclass and compare its performance against standard baselines.
CloudSim’s energy models enable evaluation of VM consolidation strategies, DVFS (Dynamic Voltage and Frequency Scaling) policies, and workload migration approaches aimed at reducing data center power consumption. This has direct relevance to the cloud industry’s growing sustainability commitments.
Simulating SLA-aware provisioning — where the cloud provider must meet committed response times, availability targets, or throughput guarantees — and evaluating how different policies affect both SLA compliance rates and infrastructure cost.
Modeling scenarios where multiple cloud providers interconnect to handle demand overflow, allowing researchers to study brokering policies, inter-cloud data transfer costs, and the QoS improvements achievable through federation.
Evaluating task offloading strategies in edge-cloud architectures, where latency-sensitive IoT workloads must be distributed optimally between edge devices and centralized cloud infrastructure.
Testing approaches to minimize the financial cost of cloud resource usage — including spot instance strategies, reserved vs. on-demand allocation mixes, and budget-constrained scheduling policies.
Evaluating how data transfer costs, network latency, and bandwidth constraints affect the performance of distributed applications — and how scheduling algorithms can be designed to minimize communication overhead.
Using WorkflowSim on top of CloudSim to model the execution of scientific workflows — dependency graphs of computational tasks typical in bioinformatics, climate modeling, and physics simulations — across cloud infrastructure.
The findings from CloudSim-based research have direct implications for how cloud platforms are designed, managed, and optimized — the same infrastructure that powers website development, digital marketing analytics, and virtually every internet-connected business service in the modern economy.
CloudSim carries zero licensing cost. It is freely downloadable, modifiable, and redistributable under the GPL license. This makes it accessible to researchers at institutions of all resource levels — from well-funded labs to individual graduate students working independently.
Unlike real cloud environments that require paid infrastructure, CloudSim runs on a standard laptop or desktop computer. A simulation that would cost hundreds or thousands of dollars on AWS can be run in minutes on personal hardware.
One of the most important advantages over real testbeds — simulations with the same parameters produce the same results every time. This reproducibility is essential for scientific validity and for comparing results across different research groups.
The Java-based object-oriented architecture allows researchers to extend nearly any component — VM schedulers, host allocators, energy models, network topologies — by subclassing existing components and overriding the relevant logic. This extensibility has powered the large ecosystem of specialized extensions.
The combination of CloudSim’s citation count, its direct extensions, and the CloudSim Plus fork has created one of the largest communities around any cloud simulation tool. Tutorials, example code, Stack Overflow threads, academic papers, and GitHub repositories are abundantly available.
The original CloudSim paper has been cited thousands of times in peer-reviewed publications. Citing CloudSim in a research paper is straightforward and well-accepted by reviewers in the cloud computing field — a practical advantage for researchers building on an established, recognized foundation.
The latest versions support simulation of IaaS, PaaS, and to some extent SaaS environments — covering the full range of cloud deployment models relevant to modern research.
Unlike simpler task scheduling simulators, CloudSim includes energy-aware and network-aware modeling out of the box — enabling multi-dimensional research that captures realistic constraints of real cloud environments.
The ability to insert new entities, pause, and resume simulations during execution enables modeling of dynamic cloud scenarios where infrastructure and workloads evolve in real time — far more realistic than static batch simulations.
A rich Javadoc API, numerous example simulation files, academic papers describing the architecture, and community-contributed tutorials provide extensive learning resources — reducing the time investment required to become productive.
CloudSim requires working knowledge of Java, object-oriented programming (including inheritance and polymorphism), and fundamental cloud computing concepts. Beginners without a solid Java background will struggle. The time investment to write your first meaningful custom simulation — beyond running the provided examples — can be substantial.
The core CloudSim framework is entirely command-line and code-based. While CloudAnalyst provides a GUI layer, it is a separate tool with its own limitations and is not part of the main CloudSim package. Researchers who prefer visual, drag-and-drop simulation configuration will find the experience frustrating.
The original CloudSim (v6.0.0-beta) is no longer under active development at the pace it once was. Bug fixes and new features are primarily advancing through CloudSim Plus rather than the original repository. Researchers starting new projects who choose the original CloudSim may encounter unresolved issues without the active support available for CloudSim Plus.
CloudSim is designed for discrete event simulation, not real-time or continuous system simulation. It is not suitable for scenarios requiring real-time reaction, live system integration, or continuous feedback loops between a running application and the simulation.
While recent versions have expanded toward PaaS and SaaS modeling, CloudSim’s architecture is fundamentally IaaS-focused — data centers, hosts, and VMs. Researchers investigating SaaS application behavior, PaaS platform dynamics, or application-level microservice architectures may find the modeling abstractions too low-level.
CloudSim is entirely Java-based. Researchers whose primary language is Python — increasingly the dominant language in both systems research and data science — must either learn Java or use a Python-to-Java bridge, adding friction to the workflow. CloudSim Plus has maintained and expanded the Java codebase but does not offer a native Python API.
While CloudSim includes network topology simulation, its network models are simplified compared to dedicated network simulators like ns-3 or OMNET++. Researchers whose primary focus is network protocol behavior or detailed traffic analysis may find CloudSim’s network layer insufficient for their needs.
Like all simulators, CloudSim models are approximations of reality. The results are only as accurate as the models and parameters used. Researchers must carefully validate their simulation parameters against real-world measurements to ensure their findings are meaningful beyond the simulation context.
The existence of multiple versions (CloudSim 3.x, CloudSim 4.x, CloudSim 5.x, CloudSim 6.x, CloudSim Plus, and various named extensions) creates confusion for newcomers trying to determine which version to use and whether existing tutorials and code examples are compatible with their chosen version.
CloudSim requires significant boilerplate Java code even for relatively simple simulations. Setting up a basic scenario with a single data center, a few VMs, and a handful of cloudlets requires dozens of lines of code — compared to more concise simulation frameworks in higher-level languages.
| Feature | CloudSim | CloudSim Plus | iFogSim | CloudAnalyst | GreenCloud |
|---|---|---|---|---|---|
| Primary Focus | IaaS cloud | IaaS + fog/edge | Fog/Edge IoT | Visual cloud analysis | Energy-efficient cloud |
| Language | Java | Java | Java | Java (GUI) | C++ (ns-2 based) |
| Free / Open Source | Yes | Yes | Yes | Yes | Yes |
| GUI | No | No | No | Yes | No |
| Python Support | No | No | No | No | No |
| Active Maintenance | Limited | Yes | Moderate | Limited | Limited |
| Network Modeling | Basic | Improved | Extended | Basic | Advanced |
| Energy Modeling | Yes | Yes | Yes | Limited | Yes (primary focus) |
| VM Scaling | Basic | Vertical + Horizontal | Basic | Basic | N/A |
| Fault Injection | No | Yes | No | No | No |
| Multi-Cloud | Yes (federation) | Yes (enhanced) | Partial | No | No |
| Community Size | Very large | Large and growing | Moderate | Moderate | Small |
| Best For | General cloud research | Modern cloud + edge | IoT/Edge research | Beginners/visual | Network + energy research |
GreenCloud (built on the ns-2 network simulator) provides more detailed network simulation and is specifically optimized for energy efficiency research. However, its C++ codebase makes it harder to extend, and its community is significantly smaller than CloudSim’s.
iFogSim is the preferred choice for fog and edge computing research — modeling the continuum from IoT edge devices through fog nodes to cloud back-ends with detailed latency and energy metrics. For cloud-only research, CloudSim remains the stronger foundation.
CloudAnalyst is ideal for educational use and initial exploration — its GUI removes the Java coding requirement entirely. For serious research requiring custom algorithms and precise control, CloudSim’s programmatic approach is necessary.
Graduate researchers and PhD students in computer science, electrical engineering, or information systems working on cloud computing, distributed systems, or related topics. The toolkit’s academic credibility, citation base, and research community make it the standard choice for cloud simulation in academic contexts.
University faculty and lecturers teaching cloud computing courses who want a hands-on simulation tool with rich example code that students can run on personal laptops without cloud account costs.
Research engineers at technology companies evaluating resource management algorithms, scheduling strategies, or SLA policies before implementation in production cloud infrastructure — where CloudSim provides a cost-free, rapid-prototyping environment.
Master’s and final-year undergraduate students completing research projects or thesis work in cloud computing — CloudSim’s documentation, examples, and community support provide a navigable path through the complexity.
Developers building cloud management software who want to validate algorithm behavior across a wide range of scenarios before deploying changes to live systems.
Industry DevOps and cloud engineers deploying and managing real workloads on AWS, Azure, or GCP. CloudSim is a research simulation tool, not a cloud management platform or monitoring solution.
Businesses looking for cloud cost optimization tools — CloudSim simulates hypothetical environments; it does not connect to real cloud APIs or optimize running infrastructure. Just as website development requires production-grade tools rather than simulators, cloud operations require live management platforms.
Python-first developers who do not want to work in Java — the tooling friction may outweigh the benefits compared to Python-based simulation or prototyping alternatives.
Researchers focused on real-time systems — CloudSim’s discrete event simulation model is not designed for real-time reaction scenarios.
Researchers needing highly accurate network simulation — for network protocol-level fidelity, dedicated network simulators (ns-3, OMNET++) provide significantly greater accuracy than CloudSim’s simplified network models.
As of 2026, CloudSim remains the most cited cloud simulation framework in the academic literature. Two active branches continue to serve the research community:
CloudSim (original) — v6.0.0-beta represents the most recent release from the original CLOUDS Lab repository. Development has slowed considerably compared to earlier years, with the original team’s primary research attention shifting to CloudSim 7G and other next-generation cloud challenges.
CloudSim Plus — The actively maintained fork continues to evolve with Java 25 compatibility, new features, regular bug fixes, and an engaged developer community. For new projects, CloudSim Plus is the recommended starting point.
The 2025 publication of CloudSim 7G — “An Integrated Toolkit for Modeling and Simulation of Future Generation Cloud Computing Environments” — signals the direction for the next chapter of CloudSim development. CloudSim 7G addresses emerging challenges in cloud simulation including heterogeneous hardware (GPUs, FPGAs), serverless computing models, and AI-driven resource management — areas not well covered by the existing framework.
The number of CloudSim-based extensions continues to grow, particularly in emerging areas:
CloudSim remains the gold standard for cloud computing simulation research in 2026 — with the important caveat that CloudSim Plus is the better choice for most new projects.
For academic researchers, students, and engineers who need to test, validate, and compare cloud resource management algorithms in a free, reproducible, and controllable environment, CloudSim (and its Plus variant) deliver an unmatched combination of capability, community support, academic credibility, and extensibility.
Its limitations are real but clearly bounded — the Java requirement, the absence of a GUI, the simplified network models, and the original version’s declining maintenance cadence are genuine trade-offs. But for the problem it was designed to solve — enabling rigorous, reproducible experimentation on cloud infrastructure and algorithms without the cost and constraints of real testbeds — CloudSim is simply the best tool available.
| Dimension | Score | Notes |
|---|---|---|
| Feature Depth | 9/10 | Comprehensive cloud modeling capabilities |
| Extensibility | 9/10 | Outstanding ecosystem of extensions |
| Ease of Setup | 7/10 | Simple setup; steep coding learning curve |
| Documentation | 8/10 | Extensive (especially CloudSim Plus) |
| Community Support | 9/10 | Large, active, well-established |
| Maintenance (Original) | 5/10 | Limited — use CloudSim Plus for new projects |
| Maintenance (Plus) | 9/10 | Actively developed and supported |
| Python Support | 2/10 | Java-only — significant barrier for some |
| GUI | 3/10 | No native GUI (CloudAnalyst is separate) |
| Real-World Accuracy | 7/10 | Good for IaaS; simplified network models |
| Cost | 10/10 | Completely free and open source |
| Overall | 8/10 | Industry-standard simulation toolkit |
What is CloudSim used for? CloudSim is used to model and simulate cloud computing environments — including data centers, virtual machines, and workloads — for the purpose of testing and evaluating resource management algorithms, scheduling policies, and provisioning strategies without using real cloud infrastructure.
Is CloudSim free? Yes. CloudSim is completely free and open source, distributed under the GPL license. There is no cost to download, use, modify, or extend it.
What programming language does CloudSim use? CloudSim is written entirely in Java. Core Java knowledge — including object-oriented programming, inheritance, and the Collections framework — is required to use it effectively.
Is CloudSim still actively maintained? The original CloudSim repository (v6.0.0-beta) has limited active development. CloudSim Plus is the actively maintained fork and is recommended for new projects. CloudSim 7G represents the next-generation research direction from the original team.
What is the difference between CloudSim and CloudSim Plus? CloudSim Plus is a completely re-engineered fork of CloudSim 3, rebuilt to address software engineering weaknesses (code duplication, poor extensibility, limited testing). It offers better accuracy, more features (including VM scaling and fault injection), Java 25 compatibility, and active maintenance.
Does CloudSim support edge computing? Not natively, but the iFogSim and EdgeCloudSim extensions — both built on the CloudSim framework — provide extensive fog and edge computing simulation capabilities.
What is a Cloudlet in CloudSim? A Cloudlet is the basic unit of computational work in CloudSim — equivalent to a job or task submitted by a user. Each Cloudlet has a defined length (in Million Instructions), required processing elements, file size, and utilization model describing how it consumes CPU, RAM, and bandwidth during execution.
Can I use CloudSim without knowing Java? Not effectively for the core framework. However, CloudAnalyst provides a GUI-based interface that does not require Java programming, and CloudSim Automation allows scenarios to be defined in YAML files. For serious research requiring custom algorithms, Java proficiency is essential.
How do I cite CloudSim in a research paper? The standard citation is: Calheiros, R.N., Ranjan, R., Beloglazov, A., De Rose, C.A.F., and Buyya, R. (2011). CloudSim: A Toolkit for Modeling and Simulation of Cloud Computing Environments and Evaluation of Resource Provisioning Algorithms. Software: Practice and Experience, 41(1), 23–50.
CloudSim helps researchers design the resource management systems that power cloud infrastructure — the same infrastructure that runs websites, search engines, digital advertising platforms, and social media at global scale. Understanding and optimizing these systems is foundational to everything the modern internet delivers.
© 2026 Macroter · Digital Marketing & SEO Services · All Rights Reserved
I’m Md Nasir Uddin, a digital marketing consultant with over 9 years of experience helping businesses grow through strategic and data-driven marketing. As the founder of Macroter, my goal is to provide businesses with innovative solutions that lead to measurable results. Therefore, I’m passionate about staying ahead of industry trends and helping businesses thrive in the digital landscape. Let’s work together to take your marketing efforts to the next level.
Paris is not simply a city. It is a global statement — a place where…
Tokyo is unlike any other city on earth — and its digital marketing landscape is…
London has always been a city that sets the pace. From the creative powerhouses of…
New York City has long been the heartbeat of global advertising and marketing. From the…
Everything you need to verify your GoDaddy website in Google Search Console — covering all…
A complete step-by-step guide covering every method to upload, link, embed, and deliver PDF files…