Software Defined Data Center Networks Research Articles

RSLB: Robust and Scalable Load Balancing in Software-Defined Data Center Networks

Data center networks demand high-performance, robust, and scalable load balancing protocols. Despite progress, existing work still cannot meet these requirements well. Software defined networking (SDN) can bring considerable flexibility to the management of data center networks. In the software-defined data center network, we design, analyze, and evaluate RSLB, a robust and scalable load balancing protocol that overcomes these challenges. RSLB uses fine-grained flowcell as the transmission unit, and uses link delay as the congestion metric. It uses a three-step routing strategy to route flowcells to the path with the least congestion. Through global congestion awareness, RSLB reduces flow completion time (FCT), and is more robust to topological asymmetries compared to existing congestion-agnostic schemes. To collect and store congestion information, RSLB adopts a distributed control structure that monitors the congestion of the entire network through multiple controllers, which makes it much more scalable for implementation in large-scale networks compare to existing congestion-aware schemes. The simulation results show that RSLB can achieve lower FCT for mice flows and higher throughput for elephant flows than existing schemes, no matter in failure-free topology or asymmetric topology.

DRL-PLink: Deep Reinforcement Learning With Private Link Approach for Mix-Flow Scheduling in Software-Defined Data-Center Networks

In datacenter networks, bandwidth-demanding elephant flows without deadline and delay-sensitive mice flows with strict deadline coexist. They compete with each other for limited network resources, and the effective scheduling of such mix-flows is extremely challenging. We propose a deep reinforcement learning with private link approach (DRL-PLink), which combines the software-defined network and deep reinforcement learning (DRL) to schedule mix-flows. DRL-PLink divides the link bandwidth and establishes some corresponding private-links for different types of flows to isolate them such that the competition among different types of flows can decrease accordingly. DRL is used to adaptively and intelligently allocate bandwidth resources for these private-links. Furthermore, to improve the scheduling policy, DRL-PLink introduces the novel clipped double Q-learning, exploration with noise, and prioritized experience replay technology for DDPG to address function approximation error, to induce lager and more randomness for exploration, as well as more effective and efficient experience replay in DRL respectively. The experiment results under actual datacenter network workloads (including Web search and data mining workload) indicate that DRL-PLink can effectively schedule mix-flows at a small system overhead. Compared with ECMP, pFabric, and Karuna, the average flow completion time of DRL-PLink decreased by 77.79%, 65.61%, and 23.34% respectively, when the deadline meet rate is increased by 16.27%, 0.02%, and 0.836% respectively. Additionally, DRL-PLink can also well achieve load balance between paths.

Intelligent routing using convolutional neural network in software-defined data center network

Intelligent routing using convolutional neural network in software-defined data center network

An Adaptive Routing Framework for Efficient Power Consumption in Software-Defined Datacenter Networks

Data Center Networks (DCNs) form the backbone of many Internet applications and services that have become necessary in daily life. Energy consumption causes both economic and environmental issues. It is reported that 10% of global energy consumption is due to ICT and network usage. Computer networking equipment is designed to accommodate network traffic; however, the level of use of the equipment is not necessarily proportional to the power consumed by it. For example, DCNs do not always run at full capacity yet the fact that they are supporting a lighter load is not mirrored by a reduction in energy consumption. DCNs have been shown to unnecessarily over-consume energy when they are not fully loaded. In this paper, we propose a new framework that reduces power consumption in software-defined DCNs. The proposed approach is composed of a new Integer Programming model and a heuristic link utility-based algorithm that strikes a balance between energy consumption and performance. We evaluate the proposed framework using an experimental platform, which consists of an optimization tool called LinGo for solving convex and non-convex optimization problems, the POX controller and the Mininet network emulator. Compared with the state-of-the-art approach, the equal cost multi-path algorithm, the results show that the proposed method reduces the power consumption by up to 10% when the network is experiencing a high traffic load and 63.3% when the traffic load is low. Based on these results, we outline how machine learning approaches could be used to further improve our approach in future work.

SEMTE: scalable and extended modular traffic engineering in software-defined data center networks

SEMTE: scalable and extended modular traffic engineering in software-defined data center networks

RETRACTED ARTICLE: Proactive Load Balancing Strategy Towards Intelligence-Enabled Software-Defined Network

Software-defined network provides a greater solution to many of the complex network management functionalities in a data center network (DCN). In a software-defined data center network, it is essential to manage heavy traffic, to avoid data loss, and to avoid server unavailability during heavy flows. One of the major tasks in a network is the load balancing in the available links. Due to the dynamic data traffic in the network, it is necessary to perform a deep learning of the long-term and the short-term data. Hence, this paper proposes an intelligent load balancing scenario that includes link load balancing, server load balancing, and traffic classification. An artificial neural network is needed because the data are uncertain all the time. Thus, our method provides improved throughput, minimized the flow completion time, and minimizes data loss.

Tennis‐resource allocation mechanism in software‐defined data center network

With the rapid development of cloud computing, most network resources are usually stored at Data Center Network (DCN) for providing the remote storage services. Under such paradigm, there arises some issues, such as traffic scheduling, energy saving and resource allocation. At present, the resource allocation in DCN has attracted the extensive concern. In order to effectively manage and allocate these resources in DCN, Software‐Defined Networking can be employed to achieve such purpose, by leveraging its natural abilities for the global control and centralized management. Another point, the tennis is a very popular sport to enhance the physical fitness. However, the tennis‐resources are considerably limited as the building of venue is quite expensive. Therefore, it becomes very significant to investigate the tennis‐resource allocation in Software‐Defined DCN (SDCN). Consider that there are many tenants using the tennis‐resources in SDCN, this paper studies the multi‐tenant tennis‐resource allocation scenario. Meanwhile, the Global Superiority Equalization algorithm is devised. At first, the max‐min fairness is used to determine the allocated number of tennis‐resources for each round. Then, the allocation state is adjusted according to the current situation until the final allocation decision is determined. The experiment results show that the proposed tennis‐resource allocation mechanism can reduce the resource allocation cost and improve the resource utilization. Accordingly, SDCN can handle more tennis‐resource requests and increase the benefit of venue provider.

DRL-R: Deep reinforcement learning approach for intelligent routing in software-defined data-center networks

Data-center networks (DCN) possess multiple new features: coexistence of elephant flow/mice flow/coflow, and coexistence of multiple network resources (bandwidth, cache and computing). The cache should be a factor of effecting routing decision because it can eliminate redundant traffic in DCN. However, the conventional routing schemes cannot learn from their previous experiences regarding network abnormalities (such as, congestion), and their metric are still the single link state (such as, hop, distance, and cost) which does not include the effect of cache. Thus, they cannot enough efficiently allocate these resources to well meet the performance requirements for various flow types. Therefore, this paper proposes deep reinforcement learning-based routing (DRL-R). Firstly, we propose a method that recombines multiple network resources with different metrics, where we recombine cache and bandwidth by quantifying their contribution score in reducing the delay. Secondly, we propose a routing scheme with resource-recombined state. By optimally allocating network resources for traffic, a DRL agent deployed on a software-defined networking (SDN) controller continually interacts with the network to adaptively perform reasonable routing according to the network state. We employ deep Q-network (DQN) and deep deterministic policy gradient (DDPG) to build the DRL-R. Finally, we demonstrate the effectiveness of DRL-R through extensive simulations. Benefitting from continuous learning with a global view, DRL-R has lower flow completion time, higher throughput and better load balance as well as better robustness, compared to OSPF. In addition, because it efficiently utilizes the network resources, DRL-R can also outperform another DRL-based routing scheme (namely TIDE). Compared to OSPF and TIDE, respectively, DRL-R can improve throughput by up to 40% and 18.5%; DRL-R can reduce flow completion time by up to 47% and 39%; DRL-R can improve the link load balance by up to 18.8% and 9.3%. Additionally, we observed that DDPG has better performance than DQN.

Fine-grained flow classification using deep learning for software defined data center networks

in a data center network, accurately classifying flow is the key to optimal schedule flow. However, the existing classification methods cannot meet the demand of real networks in terms of classification performance, detection latency, and control overhead. Thus, by combining the ability of deep learning to describe the multi-dimensional features, and the advantage of software-defined networking (SDN) in centrally controlling the network from a global viewpoint, this paper proposes a fine-grained flow classification method. This paper uses random forest technology to select eight important features in three dimensions—time distribution feature of flow, real-time feature of flow and packet header feature—for the classification model. First, this paper proposes a 2-classification scheme with a two-level architecture of pre-classification and exact-classification to detect elephant/mice flows. The pre-classification model using deep residual learning + A-Softmax with cost-sensitive is deployed on a SDN switch at the network edge to filter out a large number of mice flows. The exact-classification model using deep residual learning + AM-Softmax is deployed on the SDN controller to accurately identify the elephant flows. Second, this paper proposes a 4-classification scheme based on gated recurrent unit (GRU) to detect elephant/cheetah/tortoise/porcupine flows. Finally, the experiment results show that, when the 5th packet of a flow arrives, the 2-classification scheme can achieve a recall of up to 97.31%, an accuracy of up to 93.6%, a control overhead of 0.1kbps, and a detection latency of 7 ms. At the same time, the 4-classification scheme can achieve a recall, an accuracy, a false positive rate (FPR), a Kappa, a control overhead, and a detection latency of 83.58%, 86.53%, 5.02%, 0.797, 0.61kbps and 7.5 ms on an average, respectively. Compared with the existing methods (FlowSeer, NELLY and ESCA), all performance measures are improved to different degrees. At the same time, we also confirm the generalization of selected features and of the designed flow classification model.

REVERT: A Network Failure Recovery Method for Data Center Networks

As a repository that holds computing facilities, storage facilities, network facilities and other facilities, the Software Defined Data Center (SDDC) can provide computing and storage resources for users. For a SDDC, it is important to provide continuous services for users. Hence, in order to achieve high reliability in Software Defined Data Center Networks (SDDCNs), a network failure recovery method for software defined data center networks (REVERT) is proposed to recover failures in SDDCNs. In REVERT, the network failures that occurred in SDDCNs are classified into three types, which are switch failure, failure of links among switches and failure of links between switches and servers. Specially, except recovering the switch failure and failure of links between switches, REVERT can also recover the failures of links between the switches and servers. To achieve that, a failure preprocessing method used to classify the network failures, a data structure for storing and finding the affected flows, a server cluster agent for communicating with the server clustering algorithm and a routing path calculation method are designed in REVERT. Meanwhile, REVERT has been implemented and evaluated on RYU controller and Mininet using three routing algorithms. Compared with the link usage before recovering the network failures, when there are more than 200 flows in the network, the mean link usages only slightly increase at about 1.83 percent. More importantly, the evaluation results also demonstrate that except recovering switch failures, intra-topo link failures, REVERT has the ability of recovering failures of links between servers and edge switches successfully.

NELLY: Flow Detection Using Incremental Learning at the Server Side of SDN-Based Data Centers

The processing of big data generated by the Industrial Internet of Things (IIoT) calls for the support of processing at the edge of the network, as well as at the cloud data centers. The equal-cost multipath, which is the default routing technique in the cloud data centers, can degrade the network performance when handling mouse and elephant flows. Such degradation of performance can compromise the support of the strict quality of service requirements of the IIoT over 5G networks. Novel techniques for scheduling the elephant flows can alleviate this problem. Recently, several approaches have incorporated machine learning techniques at the controller-side in software-defined data center networks (SDDCNs) to detect elephant flows. However, these approaches can produce heavy traffic overhead, low scalability, low accuracy, and high detection time. This article introduces the Network Elephants Learner and anaLYzer (NELLY), a novel and efficient method for applying incremental learning at the server side of SDDCNs to accurately and timely identify elephant flows with low traffic overhead. Incremental learning enables NELLY to adapt to varying network traffic conditions and perform continuous learning with limited memory resources. NELLY has been extensively evaluated using real traces and various incremental learning algorithms. Results show that NELLY is accurate and supports low classification time when using adaptive decision trees algorithms.

Energy Optimization for Software-Defined Data Center Networks Based on Flow Allocation Strategies

Nowadays, energy consumption has become an important issue in data center networks. The most promising energy-saving schemes are those that shut down unnecessary network devices and links while meeting the demand of traffic loads. Existing research mainly focuses on the strategies of energy savings in software-defined data center networks (SD-DCN). Few studies have considered both energy savings and the quality of service (QoS) of the traffic load. In this paper, we investigate the energy savings guaranteed by traffic load satisfaction ratio. To ensure the minimum-power consumption in data centers, we formulate the SD-DCN energy consumption optimization problem as an Integer Linear Programming model. To achieve a high success rate for traffic transmission, we propose three flow scheduling strategies. On this foundation, we propose a strategy-based Minimum Energy Consumption (MEC) heuristic algorithm to ensure the QoS satisfaction ratio in the process of energy optimization. The results show that our algorithm can save energy efficiently under the conditions of low traffic load and medium traffic load. Under high traffic load, our algorithm can achieve better network performance than existing solutions in terms of quality of service satisfaction ratio of flow allocation.

T-DCORAL: A Threshold-Based Dynamic Controller Resource Allocation for Elastic Control Plane in Software-Defined Data Center Networks

The existing elastic control planes (ECPs) suffer from the immediacy and the computing overhead issues in software-defined data center networks (SD-DCNs). In this letter, we propose T-DCORAL which is a new ECP to mitigate both issues in SD-DCNs. T-DCORAL accelerates/decelerates the control plane by allocating a virtual CPU to controllers in runtime, whereas the existing ECPs resize the controller pool. As a result, T-DCORAL maximally reduces the latency to adjust the control plane from 46 s to 38 ms, the average CPU load by 22%, and the average rule installation time by 64.28%.

A Scalable Load Balancing Scheme for Software-Defined Datacenter Networks based on Fuzzy Logic

A Scalable Load Balancing Scheme for Software-Defined Datacenter Networks based on Fuzzy Logic

Comprehensive link sharing avoidance and switch aggregation for software-defined data center networks

An effective way to reduce network energy consumption of data center networks (DCNs) is to activate network elements as few as possible, complete transmission in as short a time as possible, and set unnecessary network elements to sleep mode. At present, most existing energy saving works considered the network energy saving from the dimension of time or power separately. However, in fact these two dimensions can interact with each other, i.e., reducing the network delay may lead to the increase of network energy consumption, and vice versa. In this paper, two dimensions of time and power are comprehensively studied in the Minimum Network Energy Consumption (MNEC) problem. First of all, we formulate the MNEC problem by considering both time and power, and prove that it is a NP-hard problem. Furthermore, we propose a heuristic Integrated Time and Power (ITP) algorithm, which combines the link sharing avoidance algorithm to reduce the network delay from the dimension of time as well as the switch aggregation algorithm to reduce the energy consumption from the power dimension. Finally, the performance of ITP algorithm is evaluated under different network topology, network size, traffic size and flow number under the network environment based on Mininet and Ryu controller. Experimental results show that the ITP algorithm outperforms the existing network energy saving algorithm in terms of energy consumption.

LLMP: Exploiting LLDP for Latency Measurement in Software-Defined Data Center Networks

The administrators of data center networks have to continually monitor path latency to detect network anomaly quickly and ensure the efficient operation of the networks. In this work, we propose Link Layer Measurement Protocol (LLMP), a prototype latency measuring framework based on the Link Layer Discovery Protocol (LLDP). LLDP is utilized by the controller to discover network topology dynamically. We insert timestamps into the optional LLDPTLV field in LLDP, so that the controller can estimate latency on any single link. The framework utilizes a reactive measurement approach without injecting any probe packets to the network. Our experiments show that the latency of a link can be measured accurately by LLMP. In relatively complex network conditions, LLMP can still maintain a high accuracy. We store the LLMP measurement results into a latency matrix, which can be used to infer the path latency.

RE-FPR: flow preemption routing scheme with redundancy elimination in Software Defined Data Center Networks

With the explosive expansion of data center sizes, a large number of the same or similar contents are requested repeatedly by the users on network edge, which causes a serious waste of network bandwidth, and further makes the network energy consumption increase remarkably. The current researches achieve energy saving by increasing network link capacity as well as eliminating the data redundancy in routers. But the existing redundancy elimination may cause the increase of router's energy consumption. To solve this problem, we propose a flow preemption routing scheme with redundancy elimination (RE-FPR). The RE-FPR scheme uses software defined networking (SDN) technology to select different routing paths for the traffic flow and control RE function on the corresponding router under two modes, i.e., traffic peak and traffic valley. Specifically, the RE-FPR scheme also employs the SDN controller to update flow states and link states of the network. We then formulate the RE-FPR problem as a power consumption minimization problem subject to flow conservation constraint and link capacity constraint. Furthermore, we solve the optimization problem by using the maximum entropy principle and propose the RE-FPR algorithm. The simulation results show that the RE-FPR algorithm outperforms the traditional flow scheduling algorithms in term of flow completion time and the number of active RE-routers/links.

Distributed Power Saving for Large-Scale Software-Defined Data Center Networks

Energy efficiency in data center networks (DCNs) is critical to the operations of modern large-scale data centers. One effective way is to make the size of DCNs elastic along with flow demands by centralized routing and scheduling, i.e., turning off idle network components to reduce the power consumption. As such, software-defined networking (SDN) is widely used for achieving such elasticity conveniently. Meanwhile, the scale and structure of modern DCNs get much larger and more complex. Central control and global computing become impractical due to the heavy time and space complexity. Therefore, distributed power control is necessary for large-scale SDN-DCNs, and yet there are few research achievements in this area. In this paper, we present an extensible energy-efficient mechanism, which: 1) leverages distributed flow routing for both intra- and inter-domain elephant flows and 2) extendedly considers distributed energy efficiency for control plane. A local power-saving function is operated within each domain of control plane, and a distributed energy-efficient routing algorithm is computed to optimize the effectiveness for the inter-domain flows. The simulation results demonstrate that this distributed mechanism applies to large-scale DCNs and achieves an effective power saving.

Online Load Balancing for Distributed Control Plane in Software-Defined Data Center Network

Distributed control plane is a common approach to improve the scalability of software-defined data center networks. However, learning how to balance the load among the controllers remains a difficult problem, since the flows in the network fluctuate frequently. In this paper, we propose an online controller load balancing (OCLB) scheme to address this issue. We first formulate the load balancing problem as an optimization problem to minimize the average controller response time. Then we decompose it into a sequence of switch migrations, with each migration aiming to reduce the average response time as much as possible based on the realtime request distribution. An OCLB algorithm is designed based on the derived optimality and termination conditions of switch migration, and is proved to be near optimal with a bounded competitive ratio. Evaluations demonstrate that our scheme can achieve near-optimal load balancing among the control plane in an online manner.

Power savings in software defined data center networks via modified hybrid genetic algorithm

AbstractIn modern data centers, power consumed by network is an observable portion of the total energy budget and thus improving the energy efficiency of data center networks (DCNs) truly matters. One effective way for this energy efficiency is to make the size of DCNs elastic along with traffic demands by flow consolidation and bandwidth scheduling, i.e., turning off unnecessary network components to reduce the power consumption. Meanwhile, having the instinct support for data center management, software defined networking (SDN) provides a paradigm to elastically control the resources of DCNs. To achieve such power savings, most of the prior efforts just adopt simple greedy heuristic to reduce computational complexity. However, due to the inherent problem of greedy algorithm, a good-enough optimization cannot be always guaranteed. To address this problem, a modified hybrid genetic algorithm (MHGA) is employed to improve the solution's accuracy, and the fine-grained routing function of SDN is fully leveraged. The simulation results show that more efficient power management can be achieved than the previous studies, by increasing about 5% of network energy savings.