Yeonho Yoo

Taeyun Roh

Youngjun Jo

Yejin Kwon

Jungyoon Shin

Geonmin Kim

System Software Laboratory (SS Lab) was established in 2023. We investigate the challenges of
system software technologies
with the goal of better
performance and efficiency
of computing infrastructures on all scales.
Our research efforts cover exciting topics like
system optimization techniques
, resource efficiency, high-performance computing, and virtualization.

Making Sense of Job Preemption for Distributed Deep Learning Acceleration

Prediction-based GPU Sharing for Distributed Training

Parameter-Efficient 12-Lead ECG Reconstruction from a Single Lead

Revisiting Traffic Splitting for Software Switch in Datacenter

Intelligent Packet Processing for Performant Containers in IoT

Prediction of the Resource Consumption of Distributed Deep Learning Systems

Harmonia: Accurate Federated Learning with All-Inclusive Dataset

Selective Preemption of Distributed Deep Learning Training

Xonar: Profiling-based Job Orderer for Distributed Deep Learning

Machine Learning-based Prediction Models for Control Traffic in SDN Systems

Bandwidth Isolation Guarantee for SDN Virtual Networks

Predictive Placement of Geo-Distributed Blockchain Nodes for Performance Guarantee

Prediction of permissioned blockchain performance

GPU Memory Prediction for Multimodal Model Training

6G Interfaces in O-RAN: Concepts, Implementation, and Future Directions

Predictive Placement of Geo-distributed Blockchain Nodes for Performance Guarantee

Prediction of Permissioned Blockchain Performance for Resource Scaling Configurations

Control Channel Isolation in SDN Virtualization: A Machine Learning Approach

TeaVisor: Network Hypervisor for Bandwidth Isolation in SDN-NV

Accurate and Efficient Monitoring for Virtualized SDN in Clouds

A Case for SDN-based Network Virtualization

Enhanced control path for repeated TCP connections

QiOi: Performance Isolation for Hyperledger Fabric

Proactive Congestion Avoidance for Distributed Deep Learning

Adaptive Control Channel Traffic Shaping for Virtualized SDN in Clouds

TensorExpress: In-Network Communication Scheduling for Distributed Deep Learning

Exploring the Characteristics of Hyperledger Fabric in Resource Consumption

Network Monitoring for SDN Virtual Networks

Resource Analysis of Blockchain Consensus Algorithms in Hyperledger Fabric

Libera for Programmable Network Virtualization

CLEO: Machine Learning for ECMP

FAVE: Bandwidth-Aware Failover in Virtualized SDN for Clouds

WhiteVisor: Support of White-Box Switch in SDN-Based Network Hypervisor

TALON: Tenant Throughput Allocation Through Traffic Load-Balancing in Virtualized Software-Defined Networks

Comprehensive Prediction Models of Control Traffic for SDN Controllers

Machine learning-based prediction models for control traffic in SDN systems

BFD-based link latency measurement in software defined networking

FlowVirt: Flow Rule Virtualization for Dynamic Scalability of Programmable Network Virtualization

LiteVisor: A Network Hypervisor to Support Flow Aggregation and Seamless Network Reconfiguration for VM Migration in Virtualized Software-Defined Networks

Proactive congestion avoidance for distributed deep learning

Accurate and efficient monitoring for virtualized SDN in clouds

Prediction of permissioned blockchain performance for resource scaling configurations

Changyong Shin

Heesang Jin

Minkoo Kang

Kyungwoon Lee

Junseok Lee

Zhixiong Niu

Younghun Go

Wonmi Choi

Peng Cheng

Yongqiang Xiong

Seungwoo Jung

Jaehyun Hwang

Gyeongsik Yang

In the blockchain network, the consensus algorithm is used to tolerate node faults with data consistency and integrity, so it is vital in all blockchain services. Previous studies on the consensus algorithm have the following limitations: 1) no resource consumption analysis was done, 2) performance analysis was not comprehensive in terms of blockchain parameters (e.g., number of orderer nodes, number of fault nodes, batch size, payload size), and 3) practical fault scenarios were not evaluated. In other words, the resource provisioning of consensus algorithms in clouds has not been addressed adequately. As many blockchain services are deployed in the form of blockchain-as-a-service (BaaS), how to provision consensus algorithms becomes a key question to be answered. This study presents a kernel-level analysis for the resource consumption and comprehensive performance evaluations of three major …

The prediction of the resource consumption for the distributed training of deep learning models is of paramount importance, as it can inform a priori users how long their training would take and also enable users to manage the cost of training. Yet, no such prediction is available for users because the resource consumption itself varies significantly according to "settings" such as GPU types and also by "workloads" like deep learning models. Previous studies have aimed to derive or model such a prediction, but they fall short of accommodating the various combinations of settings and workloads together. This study presents Driple that designs graph neural networks to predict the resource consumption of diverse workloads. Driple also designs transfer learning to extend the graph neural networks to adapt to differences in settings. The evaluation results show that Driple can effectively predict a wide range of workloads …

Current network virtualization allows tenants to have their own virtual networks. However, new demands to "program" virtual networks at a finer granularity have arisen as tenants want the ability to provision and control switches and links in their virtual networks. This study proposes a new concept called the p-NIaaS model. The p-NIaaS model enables tenants to program their own packet processing logic and monitor network status from any virtual network infrastructure, which is not possible with current network virtualization. This article presents the Libera network hypervisor that implements the p-NIaaS model. Libera overcomes the shortcomings of existing network hypervisors such as scalability, VM migration support, and flexibility. The evaluation shows that the Libera hypervisor is highly scalable and effectively supports VM migration. We also present the overheads of Libera. Libera incurs up to 11 percent …

This paper proposes V-Sight, a network monitoring framework for software-defined networking (SDN)-based virtual networks. Network virtualization with SDN (SDN-NV) makes it possible to realize programmable virtual networks; so, the technology can be beneficial to cloud services for tenants. However, to the best of our knowledge, although network monitoring is a vital prerequisite for managing and optimizing virtual networks, it has not been investigated in the context of SDN-NV. Thus, virtual networks suffer from non-isolated statistics between virtual networks, high monitoring delays, and excessive control channel consumption for gathering statistics, which critically hinders the benefits of SDN-NV. To solve these problems, V-Sight presents three key mechanisms: 1) statistics virtualization for isolated statistics, 2) transmission disaggregation for reduced transmission delay, and 3) pCollector aggregation for …

In SDN, as the control channel becomes a performance bottleneck, modeling the control channel traffic is important. Such a model is useful in predicting the control channel traffic for network provisioning. However, previously proposed models are quite limited in that they assume only the forwarding function of a specific controller for their models. To overcome the limitations, first, this paper analyzes the control traffic by seven functions (including forwarding function) of a controller. Then, we build a seven-function model to predict control channel usage and evaluate the prediction accuracy that achieves as high as 94%. Note that previous models did not have any quantitative evaluation. Our model is built with the Open Network Operating System (ONOS) controller and extended to Floodlight and POX controllers. We show that the extended model also achieves similar prediction accuracy (95%). Furthermore, we …

TensorExpress provides in-network communication scheduling for distributed deep learning (DDL). In cloud-based DDL, parameter communication over a network is a key bottleneck. Previous studies proposed tensor packet reordering approaches to reduce network blocking time. However, network contention still exists in DDL. TensorExpress mitigates network contention and reduces overall training time. It schedules tensor packets in-network using P4, a switch programming language. TensorExpress improves latency and network blocking time up to 2.5 and 2.44 times, respectively.

We introduce TeaVisor, which provides bandwidth isolation guarantee for software-defined networking (SDN)-based network virtualization (NV). SDN-NV provides topology and address virtualization while allowing flexible resource provisioning, control, and monitoring of virtual networks. However, to the best of our knowledge, the bandwidth isolation guarantee, which is essential for providing stable and reliable throughput on network services, is missing in SDN-NV. Without bandwidth isolation guarantee, tenants suffer degraded service quality and significant revenue loss. In fact, we find that the existing studies on bandwidth isolation guarantees are insufficient for SDN-NV. With SDN-NV, routing is performed by tenants, and existing studies have not addressed the overloaded link problem. To solve this problem, TeaVisor designs three components: path virtualization, bandwidth reservation, and path establishment …

As the number of tenants in clouds grows, virtualized SDN faces a challenge of control channel fairness as the control channels interfere with each other. This paper proposes an adaptive traffic shaping scheme for control channels called “Sincon.” Through experiments, Sincon achieves traffic shaping for control channels and improves the variances of control channel throughputs and forwarding setup times up to 3.8 and 2.86 times, respectively.

Virtualized software-defined networking (SDN) has been drawing increasing attention in data centers. SDN enables the creation of arbitrary and flexible virtual networks. However, given that the physical network is shared among multiple tenants, the throughput of a tenant can interfere with that of the other tenants, and the throughput requirement on each tenant cannot be met. To solve this problem, we propose TALON, a throughput allocation scheme based on traffic load-balancing. TALON allocates a throughput per tenant flow by calculating multiple paths to fulfill the traffic requirement. We design and implement TALON using an open-source network hypervisor. The evaluation results show that each tenant's throughput requirements are nearly met. In addition, the throughput increases by up to 2.29 times compared to that of a non-load-balancing network.

This article presents Elixir , an automated prediction model formulation framework for control traffic using machine learning. Control traffic is vital in software-defined networking (SDN) systems because it determines the reliability and scalability of the entire system. Various studies have sought to design control traffic prediction models for the proper provisioning and planning of SDN systems. However, previously proposed models are based on descriptive modeling, well-suited for only specific SDN system instances. Furthermore, these models exhibit poor accuracy (errors of up to 85%) because of the heterogeneity of SDN systems. Because descriptive modeling requires a significant amount of human contemplation, it is impossible to formulate adequate prediction models for countless SDN system instances. Elixir addresses this problem by applying machine learning. Elixir starts the model formulation through self …

5G networks offer various network services based on software defined networking and network function virtualization. However, certain services are sensitive to link latency which is why it is consistently observed to provide high quality services. Previous studies have proposed two approaches to this task: measuring the latency by probe packets and link-layer discovery protocol (LLDP) packets. However, they have several limitations like flow rule preconfiguration, influence of the control plane traffic, and necessity of calibration. In this paper, Bidirectional forwarding detection (BFD) based approach is proposed. The approach measures latency at the data plane with simply implemented echo mode in Open vSwitch. We evaluates and compare the proposed approach to LLDP-based one in terms of single link latency and path latency, and error rate. In addition, we verify that the control plane throughput affects link …

Network virtualization (NV) becomes an essential technology in cloud computing that isolates network flows for tenants. However, because existing NV technologies like overlay do not enable tenants to directly program (i.e., provision, control, and monitor) network resources, software-defined networking (SDN)-based NV (SDN-NV) has been proposed. Despite its great benefits, SDN-NV has been believed to bring considerable overheads due to the network hypervisor (NH). However, to date, there is no definite performance evaluation that proves the overheads of SDN-NV. To this end, this paper comprehensively investigates the performance and overheads of SDN-NV. Our experiment results reveal that SDN-NV provides the data plane performance comparable to or even better (up to 10.5× better TCP throughput) than the existing NV technologies. Also, the results on NH show that its overheads remain mostly …

We propose a new concept called "flow rule virtualization" (FlowVirt) for programmable network virtualization (P-NV). In P-NV, network hypervisor is a key component in that it plays a role in creating and managing virtual networks. This paper first reports a critical limitation of network hypervisor - scalability problem, which results in the high consumption of the switch memory, control channel, and CPU cycles: 3.9, 4.7, and 1.7 times higher than host-based network virtualization, respectively. This scalability problem arises because all the flow rules from the virtual network controllers are directly installed into switches. To resolve the scalability problem, FlowVirt introduces a flow rule abstraction: virtual and physical flow rules. By separating virtual and physical flow rules, the abstraction virtualizes flow rules so that FlowVirt can merge virtual flow rules to a smaller number of physical flow rules to be installed in switches …

Network virtualization based on software-defined networking (SDN) has become a necessary technology to provide various services in cloud datacenters. Although many network hypervisors have been proposed to support SDN-based network virtualization, their forwarding techniques excessively consume the limited ternary contents addressable memory of OpenFlow-enabled switches. Moreover, they do not consider network reconfiguration after virtual machine migration. In this paper, we propose LiteVisor, that resolves the two problems mentioned above. It develops the Locator, Identifier, and Tenant sEparating (LITE) scheme. The LITE-based forwarding enables flow aggregation that reduces switch memory consumption. In addition, LiteVisor provides seamless network reconfiguration that does not require tenant controllers to be aware of virtual machine migration based on LITE. We evaluate LiteVisor in …

This paper presents “Proactive Congestion Notification” (PCN), a congestion-avoidance technique for distributed deep learning (DDL). DDL is widely used to scale out and accelerate deep neural network training. In DDL, each worker trains a copy of the deep learning model with different training inputs and synchronizes the model gradients at the end of each iteration. However, it is well known that the network communication for synchronizing model parameters is the main bottleneck in DDL. Our key observation is that the DDL architecture makes each worker generate burst traffic every iteration, which causes network congestion and in turn degrades the throughput of DDL traffic. Based on this observation, the key idea behind PCN is to prevent potential congestion by proactively regulating the switch queue length before DDL burst traffic arrives at the switch, which prepares the switches for handling incoming DDL bursts. In our evaluation, PCN improves the throughput of DDL traffic by 72% on average.

Deep learning models have a wide spectrum of GPU execution time and memory size. When running distributed training jobs, however, their GPU execution time and memory size have not been taken into account, which leads to the high variance of job completion time (JCT). Moreover, the jobs often run into the GPU out-of-memory (OoM) problem so that the unlucky job has to restart all over. To address the problems, we propose Xonar to profile the deep learning jobs and order them in the queue. The experiments show that Xonar with TensorFlow v1.6 reduces the tail JCT by 44% with the OoM problem eliminated.

This article presents V-Sight, a network monitoring framework for programmable virtual networks in clouds. Network virtualization based on software-defined networking (SDN-NV) in clouds makes it possible to realize programmable virtual networks; consequently, this technology offers many benefits to cloud services for tenants. However, to the best of our knowledge, network monitoring, which is a prerequisite for managing and optimizing virtual networks, has not been investigated in the context of SDN-NV systems. As the first framework for network monitoring in SDN-NV, we identify three challenges: non-isolated and inaccurate statistics, high monitoring delay, and excessive control channel consumption for gathering statistics. To address these challenges, V-Sight introduces three key mechanisms: 1) statistics virtualization for isolated statistics, 2) transmission disaggregation for reduced transmission delay, and …

Performance isolation is an essential property that network virtualization must provide for clouds. This study addresses the performance isolation of the control plane in virtualized software-defined networking (SDN), which we call control channel isolation. First, we report that the control channel isolation is seriously broken in the existing network hypervisor in that the end-to-end control latency grows by up to 15 x as the number of virtual switches increases. This jeopardizes the key network operations, such as routing, in datacenters. To address this issue, we take a machine learning approach that learns from the past control traffic as time-series data. We propose a new network hypervisor, Meteor, that designs an LSTM autoencoder to predict the control traffic per virtual switch. Our evaluation results show that Meteor improves the processing latency per control message by up to 12.7x. Furthermore, Meteor reduces …

Blockchain is increasingly offered as blockchain-as-a-service (BaaS) by cloud service providers. However, configuring BaaS appropriately for optimal performance and reliability resorts to try-and-error. A key challenge is that BaaS is often perceived as a “black-box,” leading to uncertainties in performance and resource provisioning. Previous studies attempted to address this challenge; however, the impacts of both vertical and horizontal scaling remain elusive. To this end, we present machine learning-based models to predict network reliability and throughput based on scaling configurations. In our evaluation, the models exhibit prediction errors of∼ 1.9%, which is highly accurate and can be applied in the real-world.

System Software Lab. Korea University (고려대 시스템 소프트웨어 연구실)

교수님 소개

양경식

연구실 소개

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Making Sense of Job Preemption for Distributed Deep Learning Acceleration

Prediction-based GPU Sharing for Distributed Training

Making Sense of Job Preemption for Distributed Deep Learning Acceleration

연구실 보유 특허

대학별 입시끈

우리 연구실 모집

교수님 소개

양경식

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Making Sense of Job Preemption for Distributed Deep Learning Acceleration

Prediction-based GPU Sharing for Distributed Training

Making Sense of Job Preemption for Distributed Deep Learning Acceleration

연구실 보유 특허

최근 3년간 연구실적 높은 연구실

대학별 입시끈

우리 연구실 모집