Energy Efficient Ethernet Research Articles

Adaptive Frame Coalescing in Energy Efficient Ethernet With Model Predictive Control and Queuing Theory

Adaptive Frame Coalescing in Energy Efficient Ethernet With Model Predictive Control and Queuing Theory

Practical periodic strategy for 40/100 Gbps Energy Efficient Ethernet

The 40/100 Gbps Energy Efficient Ethernet (EEE) standardized by IEEE 802.3bj contains two power saving states. Correspondingly, the strategy determining the state transitions is crucial to both the power consumption and the incurred latency of frames. However, existing strategies hardly configure proper parameters for consistent good performance of power saving under variable traffic loads in reality. To address this issue, we design Practical Periodic Strategy (PPS) for the 40/100 Gbps EEE. Specifically, PPS periodically optimizes the power saving independently in each cycle. In this way, PPS decouples the incurred latency and the energy saving. (1) It limits the incurred delay by the length of the cycle. (2) It achieves good power saving by selecting a proper low power state based on the estimated traffic load in each cycle. Moreover, PPS is practical for two reasons. First, PPS involves just one cycle length parameter, which can be set according to either the given tail latency constraint or optimized automatically according to the cost function reflecting user preference. Second, the good performance of PPS is insensitive to traffic variation, because the traffic variation is smoothed by the cycles of PPS. Extensive simulations based on NS3 confirm that PPS outperforms existing strategies under different traffic patterns.

DASH Streaming traffic influence over energy efficient ethernet to improve energy savings

Dynamic Streaming over HTTP (DASH) is the main standard used in online video streaming services, given that more than 1.2 billion pay subscribers around the world use this standard. This fact entails billions of streaming connections between video servers and client displays. These devices involved in the streaming connection use an Ethernet Interface Card that consumes energy. In order to reduce the energy consumption, IEEE proposed the 802.3az Energy Efficient Ethernet standard, with a mechanism to make the network card change to a low power consumption mode when it is not in transmission mode. This behavior will be beneficial for services where traffic is sent in bursts, for instance video packet bursts like in video streaming over Real Time Transport Protocol in IPTV or the widely used DASH standard. Therefore, in this study the Ethernet traffic pattern when transmitting online video content using DASH is characterized in order to analyze the efficiency of the IEEE 802.3az standard under this video streaming scenario, and to verify the convenience of activating this energy saving alternative at the network interface of billions of client devices. The experiments have been conducted using a test-bed consisting of a full DASH streaming architecture, comparing different video segment sizes and changing the available bandwidth during the experiments in different scenarios in order to analyze the effect of the DASH content segment size on the Ethernet traffic pattern to identify the trade-off between energy efficiency, the energy savings, and the impact on the performance of the dynamic adaptation on the video streaming and reproduction.

Modeling and Analysis of Latency Distribution in the 40-100Gbps Dual-Mode Energy Efficient Ethernet

Recently, the 40-100Gbps Energy Efficient Ethernet (EEE) is standardized by IEEE 802.3bj. Unlike the 1-10Gbps EEE, the 40-100Gbps EEE includes two energy-saving states with different power consumption and state transition times. The default Dual-Mode strategy is the fundamental policy, which utilizes these states to make a trade-off between the energy saving and the incurred latency of frames. Although numerous analytical works have been proposed for EEE strategies, most of them can only be applied to the single energy-saving state condition of 1-10Gbps EEE, or can only capture the mean latency of frames. However, the interactive applications always place a soft or hard deadline on frame transmission. Accordingly, the tail latency of frames becomes crucial. Therefore, we model and analyze the latency distribution of frames in the 40-100Gbps Dual-Mode EEE in this work. The guidelines provided by analytical results help the Dual-Mode strategy to fulfill the given tail latency requirement and make a better trade-off between the energy saving and the incurred latency under different traffic loads. In addition, this model is powerful as it can generate the major results of several existing works under the corresponding simplified scenarios. Finally, the analytical results are verified by trace-driven simulations based on NS3.

Generalized Predictive Control of the Time Window for Energy Efficient Ethernet With Prediction

The strategy of determining the occasion to enter and leave the Low Power Idle (LPI) state is crucial to the power savings and incurred queuing delays of the Energy Efficient Ethernet. The novel strategy EEEP (Energy Efficient Ethernet with Prediction) proposed recently achieves good energy-saving and limited latency by allocating the LPI duration within a fixed time window. In this letter, we observe that the average incurred delays increase linearly while the energy consumption ratio curve is hyperbolic with the increase of the time window parameter. Accordingly, when the performance is quantified by the cost defined as a linear combination of power consumption ratio and average queuing delay, an optimal time window parameter would minimize the cost. Furthermore, we propose the GPC-EEEP strategy, which automatically adjusts the time window parameter of EEEP to the optimal value with the generalized predictive control method. Simulations driven by both synthetic traffic and real traces confirm that GPC-EEEP significantly improves the cost by automatically settling the time window parameter according to dynamic traffic.

M/M/1 model of Energy-Efficient Ethernet with byte-based coalescing

Communication networks have been recognized as substantial energy consumer. However, the ubiquity of Ethernet links provides opportunity for energy savings with Energy-Efficient Ethernet standard and packet coalescing. So far, theoretical analysis of coalescing algorithms for Energy-Efficient Ethernet has assumed that the coalescing limit is expressed in packets; however, as Ethernet links are byte-congestible resources in nature, we argue that byte-based coalescing algorithms should be employed. To that goal, we propose an M/M/1 model for byte-based coalescing on Ethernet links compliant to Energy-Efficient Ethernet standard. The model is based on compound Poisson distribution and provides single formula for calculation of expected low-power state duration and, in turn, of energy savings. The model is applicable to 10GBASE-T Ethernet and to emerging 2.5GBASE-T and 5GBASE-T standards. Detailed simulation results show a good match of achieved energy savings to those predicted by the model. The paper discusses application of the proposed model for the evaluation of energy efficiency of 10GBASE-T Ethernet links in future IoT data centers within Smart Cities.

Leveraging energy saving capabilities of current EEE interfaces via pre-coalescing

The low power idle mode implemented by Energy Efficient Ethernet (EEE) allows network interfaces to save up to 90% of their nominal energy consumption when idling. There is an ample body of research that recommends the use of frame coalescing algorithms—that enter the low power mode as soon as there is no more traffic waiting to be sent, and delay the exit from this mode until there is an acceptable amount of traffic queued—to minimize energy usage while maintaining an acceptable performance. However, EEE capable hardware from several manufactures delays the entrance to the low power mode for a considerable amount of time (hysteresis). In this paper we augment existing EEE energy models to account for the hysteresis delay and show that, using the configuration ranges provided by manufacturers, most existing EEE networking devices are unable to obtain significant energy savings. To improve their energy efficiency, we propose to implement frame coalescing directly at traffic sources, before reaching the network interface. We also derive the optimum coalescing parameters to obtain a given target energy consumption at the EEE device when its configuration parameters are known in advance.

Power Efficiency and Delay Tradeoff of 100G Energy Efficient Ethernet Protocol

This paper investigates the dual-mode power-saving strategy designed for the 100G Energy Efficient Ethernet. The process of this strategy is a sequence of cycles, where each cycle is an interval elapsed between two consecutive instants when the buffer of the Ethernet interface becomes empty. In each cycle, the interface first enters the fast-wake mode to perform the conditional sleep-mode selection of the strategy. If the number of arrivals during the fast-wake mode reaches a threshold, the interface directly wakes up; otherwise, it proceeds to the deep-sleep mode. The sequence of cycles switches between two types: deep-sleep cycles with deep-sleep mode and light-sleep cycles without deep-sleep mode. We analyze the dual-mode strategy based on the condition of the number of arrivals during the fast-wake mode. We first derive the weights of conditions according to the feature of the conditional sleep-mode operation, and then calculate the unconditioned performance measure of the dual-mode strategy based on the weighted average of that of these two kinds of cycles. Finally, we obtain the close-form expressions of the power efficiency and the mean delay, based on which we provide a set of parameter selection rules. We show that the dual-mode strategy with these rules can select suitable sleep modes according to the instantaneous traffic rate, and thus perform well under bursty input traffic.

Dynamic EEE Coalescing: Techniques and Bounds

Frame coalescing is one of the most efficient techniques to manage the low power idle (LPI) mode supported by Energy Efficient Ethernet (EEE) interfaces. This technique enables EEE interfaces to remain in the LPI mode for a certain amount of time upon the arrival of the first frame (time-based coalescing) or until a predefined amount of traffic accumulates in the transmission buffer (size-based coalescing). This paper provides new insights on the practical efficiency limits of both coalescing techniques. In particular, we derive the fundamental limits on the maximum energy savings considering a target average frame delay. Additionally, we present new open-loop adaptive variants of both time-based and size-based coalescing techniques. These proposals dynamically adjust the length of the sleeping periods in accordance with actual traffic conditions to reduce energy consumption while keeping the average delay near a predefined value simultaneously. Analytical and simulation results show that the energy consumption of both proposals is comparable to the fundamental limits. Consequently, we recommend the usage of the time-based algorithm in most scenarios because of its simplicity as well as its ability to bound the maximum frame delay at the same time.

Improvement of the Prediction-Based Energy Efficient Ethernet Strategy

Ethernet consumes maximum energy even when there is no data transmission. To reduce the power consumption, IEEE 802.3az standardizes the Energy Efficient Ethernet that enhances Ethernet with the low power idle state without data transmission. However, this standard does not describe the specific strategy about when the Ethernet link will enter or exit the low power idle state. Recently, they proposed the EEEP strategy for the 1–10Gbps EEE to reduce power consumption. Specifically, EEEP predicts the future traffic in a time window by the Autoregressive Integrated Moving Average (ARIMA) model and determines when to enter or exit the low power idle state according to the prediction results. However, the EEEP strategy relies on the prediction accuracy of the ARIMA model for good energy saving. This paper proposes to use the Long Short Term Memory (LSTM) model for EEEP to improve the prediction accuracy. Owning to the historic traffic information, the LSTM model can achieve about 11% improvement on the accuracy compared to ARIMA, and thus helps EEEP to achieve better energy saving, according to our trace-driven simulation results.

POWAR

In order to save energy in HPC interconnection networks, one usual proposal is to switch idle links into a low-power mode after a certain time without any transmission, as IEEE Energy Efficient Ethernet standard proposes. Extending the low-power mode mechanism, we propose POW er- A ware R outing ( POWAR ), a simple power-aware routing and selection function for fat-tree and torus networks. POWAR adapts the amount of network links that can be used, taking into account the network load, and obtaining great energy savings in the network (55%--65%) and the entire system (9%--10%) with negligible performance overhead.

Matching SDN and Legacy Networking Hardware for Energy Efficiency and Bounded Delay.

Both economic and environmental costs are driving much research in the area of the energy efficiency of networking equipment. This research has produced a great amount of proposals. However, the majority of them remain unimplemented due to the lack of flexibility of current hardware devices and a certain lack of enthusiasm from commercial vendors. At the same time, Software-Defined Networking (SDN) has allowed customers to control switching decisions with a flexibility and precision previously unheard of. This paper explores the potential convergence between the two aforementioned trends and presents a promising power saving algorithm that can be implemented using standard SDN capabilities of current switches, reducing operation costs on both data centers and wired access networks. In particular, we focus on minimizing the energy consumption in bundles of energy-efficient Ethernet links leveraging SDN. For this, we build on an existing theoretical algorithm and adapt it for implementing with an SDN solution. We study several approaches and compare the resulting algorithms not only according to their energy efficiency, but also taking into account additional QoS metrics. The results show that the resulting algorithm is able to closely match the theoretical results, even when taking into account the requirements of delay-sensitive traffic.

Can Energy-Aware Routing Improve the Energy Savings of Energy-Efficient Ethernet?

Telecommunication networks are over-provisioned with redundant resources in order to cope with traffic load during peak hours and to quickly recover from failures. However, much of the resources are underutilized during long periods of time, but still consuming full energy. With the growing concerns of energy waste and greenhouse gas emissions, the network design principles tend to shift towards allocation of resources on-demand for energy-efficiency. In this paper, we analyze and evaluate the performance of two different energy-saving techniques, namely energy saving topology control (ESTOP) and energy-efficient Ethernet (EEE). We investigate the energy-saving characteristics of ESTOP+EEE; the combination of ESTOP and EEE. The evaluation is conducted in OMNet++ with realistic and synthetic network topologies under varying traffic conditions. The results indicate that the combination has a significant potential for saving energy, compared to running ESTOP or EEE alone, but that the amount of energy savings depends on topology, traffic load, and the chosen target connectivity level for ESTOP. In particular, the results show that the target connectivity level needs to be carefully matched to the topology and the current traffic situation, suggesting that ESTOP+EEE would be suitable where the target connectivity level is dynamically adjusted according to traffic variations.

PerfBound: Conserving Energy with Bounded Overheads in On/Off-Based HPC Interconnects

Energy and power are key challenges in high-performance computing. System energy efficiency must be significantly improved, and this requires greater efficiency in all subcomponents. An important target of optimization is the interconnect, since network links are always on, consuming power even during idle periods. A large number of HPC machines have a primary interconnect based on Ethernet (about 40 percent of TOP500 machines), which, since 2010, has included support for saving power via Energy Efficient Ethernet (EEE). Nevertheless, it is unlikely that HPC interconnects would use these energy saving modes unless the performance overhead is known and small. This paper presents PerfBound, a self-contained technique to manage on/off-based networks such as EEE, minimizing interconnect link energy consumption subject to a bound on the performance degradation. PerfBound does not require changes to the applications and it uses only local information already available at switches and NICs without introducing additional communication messages, and is also compatible with multi-hop networks. PerfBound is evaluated using traces from a production supercomputer. For twelve out of fourteen applications, PerfBound has high energy savings, up to 70 percent for only 1 percent performance degradation. This paper also presents DynamicFastwake, which extends PerfBound to exploit multiple low-power states. DynamicFastwake achieves an energy–delay product 10 percent lower than the original PerfBound technique.

Optimum Traffic Allocation in Bundled Energy-Efficient Ethernet Links

The energy demands of Ethernet links have been an active focus of research in the recent years. This work has enabled a new generation of Energy Efficient Ethernet (EEE) interfaces able to adapt their power consumption to the actual traffic demands, thus yielding significant energy savings. With the energy consumption of single network connections being a solved problem, in this paper we focus on the energy demands of link aggregates that are commonly used to increase the capacity of a network connection. We build on known energy models of single EEE links to derive the energy demands of the whole aggregate as a function on how the traffic load is spread among its powered links. We then provide a practical method to share the load that minimizes overall energy consumption with controlled packet delay, and prove that it is valid for a wide range of EEE links. Finally, we validate our method with both synthetic and real traffic traces captured in Internet backbones.

Dual-mode Energy Efficient Ethernet with Packet Coalescing: Analysis and simulation

The recently-proposed and studied Dual-Mode Energy Efficient Ethernet (EEE) amends the conventional EEE with an additional low-power mode called Fast Wake. Dual-Mode EEE has shown to be much more effective than EEE in reducing the power consumption of Ethernet links at 40Gb/s and higher due to impractically-high transition times of using only one low-power mode at these capacities. However, even with two low-power modes, the still relatively-high transition times between the modes prevent Dual-Mode EEE from taking full advantage of the power reduction of the low-power modes and approach energy-proportional operation.We design and formulate Packet Coalescing for Dual-Mode EEE links at 40Gb/s and higher. Packet Coalescing is to collect multiple packets before transmitting them on a link as a burst of back-to-back packets. By coalescing packets into bursts, the overhead of transition times can be reduced and nearly energy-proportional operation can be achieved, albeit at the expense of a bounded increase in the packet delay. We analytically model the power consumption of Dual-Mode EEE with Packet Coalescing and present closed-form formulae for the average percentage of time a link would spend in each low-power mode and transition. We also simulate a 40Gb/s Dual-Mode EEE link with Packet Coalescing and present results suggesting that the compromise in the average per-packet delay is likely to be worthwhile under most circumstances considering the power reduction.

Energy Efficient Ethernet on MapReduce Clusters: Packet Coalescing To Improve 10GbE Links

An important challenge of modern data centers is to reduce energy consumption, of which a substantial proportion is due to the network. Switches and NICs supporting the recent energy efficient Ethernet (EEE) standard are now available, but current practice is to disable EEE in production use, since its effect on real world application performance is poorly understood. This paper contributes to this discussion by analyzing the impact of EEE on MapReduce workloads, in terms of performance overheads and energy savings. MapReduce is the central programming model of Apache Hadoop, one of the most widely used application frameworks in modern data centers. We find that, while 1GbE links (edge links) achieve good energy savings using the standard EEE implementation, optimum energy savings in the 10 GbE links (aggregation and core links) are only possible, if these links employ packet coalescing. Packet coalescing must, however, be carefully configured in order to avoid excessive performance degradation. With our new analysis of how the static parameters of packet coalescing perform under different cluster loads, we were able to cover both idle and heavy load periods that can exist on this type of environment. Finally, we evaluate our recommendation for packet coalescing for 10 GbE links using the energy-delay metric. This paper is an extension of our previous work [1] , which was published in the Proceedings of the 40th Annual IEEE Conference on Local Computer Networks (LCN 2015).

Power Efficiency and Delay Tradeoff of 10GBase-T Energy Efficient Ethernet Protocol

In this paper, we study the power efficiency and delay performance of the burst mode transmission (BTR) strategy for the IEEE 802.3az energy efficient Ethernet (EEE) protocol. In the BTR strategy, the Ethernet interface goes to sleep once its transmission buffer becomes empty and wakes up as soon as the first arrival has waited for time $\tau $ or the $N$ -th frame arrives at the interface. Based on the number of arrivals during the vacation time, a new approach is proposed to analyze the M/G/1 queue with vacation times that are governed by the arrival process and the $\tau $ and $N$ parameters of BTR strategy. Our key idea is to establish the connection between the vacation time and the arrival process to account for their dependency. We first derive the distribution of the number of arrivals during a vacation time based on an event tree of the BTR strategy, from which, we obtain the mean vacation time and the power efficiency. Next, from the condition on the number of arrivals at the end of a vacation period, we derive a generalized P-K formula of the mean delay for EEE systems, and prove that the classical P-K formula of the vacation model is only a special case when the vacation time is independent of the arrival process. Our analysis demonstrates that the $\tau $ policy and $N$ policy of the BTR strategy are compensating each other. The $\tau $ policy ensures the frame delay is bounded when the traffic load is light, while the $N$ policy ensures the queue length at the end of vacation times is bounded when the traffic load is heavy. These results, in turn, provide the rules to select appropriate $\tau $ and $N$ . Our analytical results are confirmed by simulations.

Delay Properties of Energy Efficient Ethernet Networks

Networking operational costs and environmental concerns have lately driven the quest for energy efficient equipment. In wired networks, energy efficient Ethernet (EEE) interfaces can greatly reduce power demands when compared to regular Ethernet interfaces. Their power saving capabilities have been studied and modeled in many research articles in the last few years, together with their effects on traffic delay. However, to this date, all articles have considered them in isolation instead of as part of a network of EEE interfaces. In this paper we develop a model for the traffic delay on a network of EEE interfaces. We prove that, whatever the network topology, the per interface delay increment due to the power savings capabilities is bounded and, in most scenarios, negligible. This confirms that EEE interfaces can be used in all but the most delay constrained scenarios to save considerable amounts of power.

Modeling and understanding burst transmission for energy efficient ethernet

Recently, the energy consumption of Ethernet has become one of the hottest topics focused by both academic committee and industry, especially with the increase of the link speed from 1Gbps to 10/40Gbps nowadays or even 100/400Gbps in the near future. To save the energy consumed by the Ethernet, the Energy Efficient Ethernet (EEE) is developed and standardized by the IEEE 802.3az work group. When there is no incoming traffic, the EEE can saves 90% of its energy consumption by entering into the Low Power Idle (LPI) mode. To maximize the energy saving of Ethernet, the Burst TRansmission (BTR) mechanism, which defines a new way to utilize the LPI mode, is developed as a mechanism for EEE. Prior work theoretically shows that the BTR mechanism makes a tradeoff between the energy saving and the queuing delay. However, the traffic pattern, on which the performance of EEE greatly depends, is assumed to be deterministic in their analyses. Besides, their models made estimation for many situations. In this paper, we model EEE with the BTR mechanism and provide analytical understanding on the BTR mechanism. We propose two actual models: one focuses on the buffer size limit, the other concentrates on tolerable packet delay additionally. We draw some guidelines of parameter selection and policies design for EEE from combination of theory conclusions and simulation results. The results show that the saved energy can be constrained by link occupancy even though the buffer size is variational. The buffer full triggered wake-up policy can achieve ideal ratio of energy consumption and arrival rate within the scope of the buffer as well. However, the tolerable delay cannot be guaranteed by any policies. The buffer size is even fixed, which affects the flexibility of demanded delay for different business. The policy considering tolerable delay is supposed to be a little better than the other policy, with a little more complicated design. Thus we design an adaptive mechanism both in quantitative and qualitative: detect the load utilization, apply the buffer full triggered wake-up policy for higher load utilization link, while applying the buffer full and timeout triggered wake-up policy for the delay sensitive business and tiny arrival rate.