NUMFabric

Nagaraj, Kanthi; Bharadia, Dinesh; Mao, Hongzi; Chinchali, Sandeep; Alizadeh, Mohammad; Katti, Sachin

doi:10.1145/2934872.2934890

Cited by 72 publications

(6 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We observe that serving both links with maximum rates equal to the link capacity does not lead to minimum delay in general cases, which validates the necessity of refined control of link rates based on (9). This result also gives insights on the demand-aware bandwidth allocation in data center networks [52], [63], [64], where allocating bandwidth in proportion to the demands from different ingress nodes leads to minimum delay when overload occurs. The minimum total bandwidth c 1 +c 2 required to achieve the global minimum delay as in the case with unlimited capacity is µ, where c i = λi λ1+λ2 µ, i = 1, 2.…”

Section: A N × 1 Networksupporting

confidence: 54%

Queueing Delay Minimization in Overloaded Networks via Rate Control

Modiano

2022

2022 58th Annual Allerton Conference on Communication, Control, and Computing (Allerton)

View full text Add to dashboard Cite

We develop link rate control policies to minimize the queueing delay of packets in overloaded networks. We show that increasing link rates does not guarantee delay reduction during overload. We consider a fluid queueing model that facilitates explicit characterization of the queueing delay of packets, and establish explicit conditions on link rates that can minimize the average and maximum queueing delay in both single-hop and multi-stage (switching) networks. These min-delay conditions require maintaining an identical ratio between the ingress and egress rates of different nodes at the same layer of the network. We term the policies that follow these conditions rate-proportional policies. We further generalize the rate-proportional policies to queue-proportional policies, which minimize the queueing delay asymptotically based on the time-varying queue length while remaining agnostic of packet arrival rates. We validate that the proposed policies lead to minimum queueing delay under various network topologies and settings, compared with benchmarks including the backpressure policy that maximizes network throughput and the max-link-rate policy that fully utilizes bandwidth. We further remark that the explicit min-delay policy design in multi-stage networks facilitates co-optimization with other metrics, such as minimizing total bandwidth, balancing link utilization and node buffer usage. This demonstrates the wider utility of our main results in data center network optimization in practice.

show abstract

Section: A N × 1 Networksupporting

confidence: 54%

Queueing Delay Minimization in Overloaded Networks via Rate Control

Modiano

2022

2022 58th Annual Allerton Conference on Communication, Control, and Computing (Allerton)

View full text Add to dashboard Cite

show abstract

“…KAIST [18] is a real trajectory dataset of user mobility, which forms a mobile adhoc network. NumFab is a simulated host server flow dataset of DCN generated via the simulation code of [19]. According to the second strategy in Section II, I constructed snapshots based on the (i) distance between each UE pair for KAIST and (ii) traffic volume between each host pair for NumFabric.…”

Section: Extensive Experiments a Datasets Of Various Scenariosmentioning

confidence: 99%

High-Quality Temporal Link Prediction for Weighted Dynamic Graphs via Inductive Embedding Aggregation

Meng

Zhang

Bai

et al. 2023

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Network dynamic (e.g., traffic burst in data center networks and channel fading in cellular WiFi networks) has a great impact on the performance of communication networks (e.g., throughput, capacity, delay, and jitter). This article proposes a unified prediction-based method to handle the dynamic of various network systems. From the view of graph deep learning, I generally formulate the dynamic prediction of networks as a temporal link prediction task and analyze the possible challenges of the prediction of weighted networks, where link weights have the wide-value-range and sparsity issues. Inspired by the highresolution video frame prediction with generative adversarial network (GAN), I try to adopt adversarial learning to generate high-quality predicted snapshots for network dynamic, which is expected to support the precise and fine-grained network control. A novel high-quality temporal link prediction (HQ-TLP) model with GAN is then developed to illustrate the potential of my basic idea. Extensive experiments for various application scenarios further demonstrate the powerful capability of HQ-TLP.

show abstract

“…In summary, existing studies did not learn from their previous experiences regarding network abnormalities such as congestion. Furthermore, they could not effectively address network dynamics including time‐varying resource state and user demands .…”

Section: Introductionmentioning

confidence: 98%

“…In widely deployed software‐defined data‐center network (SD‐DCN), research has determined that there are typically three types of flow: mice flows (MF), elephant flows with known sizes (EF1), and elephant flows with unknown sizes (EF2). MF are short occasional flows of data, typically generated by latency‐sensitive applications.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Mix‐flow scheduling using deep reinforcement learning for software‐defined data‐center networks

Liu¹,

Cai

Wang

et al. 2019

Internet Technology Letters

View full text Add to dashboard Cite

For a mix‐flow scenario in software‐defined data‐center networks, how to simultaneously achieve the different performance requirements of the different types of flows is a considerable challenge. This paper proposes a mix‐flow scheduling scheme based on deep reinforcement learning (DRL). This paper establishes three private link sets for three types of flows. Then, DRL is employed to adaptively and intelligently allocate bandwidth for each type of flow according to the traffic variations across time and space. A novel metric is designed as a function of DRL's reward to guide the process of simultaneously maximizing the deadline meet rate for mice flows (MF) and minimizing the flow completion time for elephant flows. Within these three private link sets, three flow‐scheduling strategies (ie, priority‐based allocation for MF, stable matching‐based allocation for elephant flows with unknown sizes, and proportion‐based allocation for elephant flows with known sizes) are employed. A simulation demonstrates the effectiveness of the proposed scheme compared with previous methods (Fincher and pFabric). DRL‐Flow's overhead also is minimal to satisfy the scalability well and is deployable in a large‐scale network.

show abstract

NUMFabric

Cited by 72 publications

References 75 publications

Queueing Delay Minimization in Overloaded Networks via Rate Control

Queueing Delay Minimization in Overloaded Networks via Rate Control

High-Quality Temporal Link Prediction for Weighted Dynamic Graphs via Inductive Embedding Aggregation

Mix‐flow scheduling using deep reinforcement learning for software‐defined data‐center networks

Contact Info

Product

Resources

About