A port-based forwarding load-balancing scheduling approach for cloud datacenter networks

Abstract

Today’s datacenter networks (DCNs) scale is rapidly increasing because of the wide deployment of cloud services and the rapid rise of edge computing. The bandwidth consumption and cost of a DCN are growing sharply with the extensions of network size. Thus, how to keep the traffic balanced is a key and challenging issue. However, the traditional load balancing algorithms such as Equal-Cost Multi-Path routing (ECMP) are not suitable for high dynamic traffic in cloud DCNs. In this paper, we propose a port-based forwarding load balancing scheduling (PFLBS) approach for Fat-tree based DCNs with some new features which can overcome the disadvantages of the existing load balancing methods in the following aspects. Firstly, we define a port-based source-routing addressing scheme, which decreases the switch complexity and makes the table-lookup operation unnecessary. Secondly, based on this addressing scheme, we proposed an effective routing mechanism which can obtain multiple available paths for flow scheduling based in Fat-tree. All the path information is saved in servers and each server only needs to maintain its own path information. Thirdly, we propose an efficient algorithm to implement large flows scheduling dynamically in terms of current link utilization ratio. This method is suitable for cloud DCNs and edge computing, which can reduce the complexity of the switches and the power consumption of the whole network. The experiment results indicate that the PFLBS approach has better performance compared with the ECMP, Hedera and MPTCP approaches, which decreases the flow completion time and improves the average throughput significantly. PFLBS is simple and can be implemented with a few signaling overheads.

FMap: A fuzzy map for scheduling elephant flows through jumping traveling salesman problem variant toward software‐defined networking‐based data center networks

“…6 Although Hedera suffers from several serious inefficiencies, it is one of the well-known methods for EF scheduling which is still being compared with the state-of-the-art works (due to its remarkable framework within holistic traffic patterns considerations and DCN's specifics). 18,28,34 • ECMP 22 is a traffic scheduling scheme that most DCNs use as a routing approach that carries flows along equivalent paths. It splits the traffic of the same destination through multiple next-hops with equal costs.…”

“…Therefore, to MFs, even a processed and fine-grained solution may lead to nothing except an increase in MF's FCT and bottleneck intensification. Furthermore, ECMP, which performs static flow stripping across the available paths, acts well for most of MFs, 34 and it has been utilized for delivering MFs in other works within EF scheduling concentration as well. 12,28 Thus, an indirect plan which emphasizes EFs (which have the potential to withstand little computation time for being served by demanded throughput) might have significantly better impacts on MFs.…”

“…However, if there exist multiple paths, the chosen route might not be optimal in terms of bandwidth utilization, leading to an unbalanced distribution of network traffic 6 . Although Hedera suffers from several serious inefficiencies, it is one of the well‐known methods for EF scheduling which is still being compared with the state‐of‐the‐art works (due to its remarkable framework within holistic traffic patterns considerations and DCN's specifics) 18,28,34 ECMP 22 is a traffic scheduling scheme that most DCNs use as a routing approach that carries flows along equivalent paths.…”

QoS Analysis for Cloud-Based IoT Data Using Multicriteria-Based Optimization Approach

“…It is obvious that this version of the genetic algorithm solves the problem of the allocation of very simplified services compared to real cloud services. However, this version of simplified service allocation allows analysis to be focused on optimizing the various selected QoS metrics [38][39][40][41][42]. Indeed, using this genetic algorithm as a placement meta-heuristic makes it possible to evaluate both the quality of its optimization but also to demonstrate the impact of metrics on top of each other, highlighting the advantage of using a multicriteria approach for service allocation purposes.…”

DeepRLB: A deep reinforcement learning‐based load balancing in data center networks

“…Then, the port with the lowest current queue occupancy is chosen from d + m output ports, and the switch sends the packet to this port. Liu et al 28 proposed PFLBS as a port‐based forwarding load‐balancing scheduling scheme for Fat‐tree data center topology. They defined a port‐based source‐routing addressing (PA) mechanism for forwarding addresses and designed a routing approach to obtain multiple available paths based on PAs.…”