Related Experiment Videos
Performance evaluation of TCP congestion control variants across application workloads in cloud based networks
Aravinda Swamy D1, Ananya Hegde1, Anantha Chary C M1
1Department of Information Science and Engineering, B.M.S. College of Engineering, Bengaluru, India.
Scientific Reports
|July 10, 2026
Summary
Selecting the optimal Transmission Control Protocol (TCP) congestion control algorithm is crucial for shared cloud environments. Our study compares Cubic, Reno, and Bottleneck Bandwidth and Round-trip propagation time (BBR) across diverse workloads, revealing significant performance differences.
Area of Science:
- Computer Science
- Network Engineering
- Cloud Computing
Background:
- Shared cloud environments necessitate careful selection of Transmission Control Protocol (TCP) congestion control algorithms.
- Limited head-to-head comparisons exist for widely used algorithms across heterogeneous workloads.
Purpose of the Study:
- To evaluate and compare the performance of three prominent TCP congestion control algorithms: Cubic, Reno, and Bottleneck Bandwidth and Round-trip propagation time (BBR).
- To provide workload-specific guidance for algorithm selection in multitenant cloud deployments.
Main Methods:
- Experiments were conducted using four Amazon Web Services (AWS) EC2 instances in a dumbbell network topology.
- Evaluated algorithms under four distinct workload types: synthetic throughput (iperf3), real-time stream processing (Kafka, Flink), distributed I/O, and compute-intensive sorting (Hadoop TeraSort).
- Performance metrics included throughput, end-to-end latency, retransmission count, and job completion time.
Main Results:
- Cubic, Reno, and BBR exhibited distinct performance characteristics that varied significantly based on the workload type.
- Differences were observed across all measured metrics: throughput, latency, retransmissions, and job completion times.
- Workload-specific performance variations highlight the importance of tailored algorithm selection.
Conclusions:
- The choice of TCP congestion control algorithm critically impacts performance in shared cloud environments.
- Results offer concrete, workload-dependent recommendations for cloud operators selecting congestion control policies.
- Understanding these performance differences is essential for optimizing cloud resource utilization and application performance.
Related Concept Videos
01:23Maximum Power Flow and Line Loadability
The maximum power flow for lossy transmission lines is derived using ABCD parameters in phasor form. These parameters create a matrix relationship between the sending-end and receiving-end voltages and currents, allowing the determination of the receiving-end current. This relationship facilitates calculating the complex power delivered to the receiving end, from which real and reactive power components are derived.
01:01Production Efficiency
Net production efficiency (NPE) is the efficiency at which organisms assimilate energy into biomass for the next trophic level. Due to low metabolic rates and less energy spent on thermoregulatory processes, the NPE of ectotherms (cold-blooded animals) is 10 times higher than endotherms (warm-blooded animals).
02:30Quantifying Work
As a system undergoes a change, its internal energy can change, and energy can be transferred from the system to the surroundings, or from the surroundings to the system.
01:28Pipe Flowrate Measurement
In pipe flow measurement, orifice, nozzle, and Venturi meters are commonly used to determine fluid flowrates by constricting the flow area, which increases fluid velocity and reduces pressure. This pressure difference, governed by Bernoulli's principle and adjusted for real-world conditions, is essential for calculating flowrate. Each meter type is suited to specific applications based on accuracy, efficiency, and compatibility with various flow conditions.
The orifice meter is a simple,...
The orifice meter is a simple,...
01:19Distributed Loads
Distributed loads are a common type of load that engineers and scientists encounter in various practical situations. Distributed loads often refer to a type of load spread over a surface or a structure and can be modeled as continuous force per unit area.
For example, consider a bookshelf filled with books stacked vertically adjacent to each other. The weight of the books is evenly distributed over the length of the shelf. As a result, the pressure at different locations on the surface of the...
For example, consider a bookshelf filled with books stacked vertically adjacent to each other. The weight of the books is evenly distributed over the length of the shelf. As a result, the pressure at different locations on the surface of the...