Papers by Mohammad Noormohammadpour
IEEE Communications Letters, 2018
—Long flows contribute huge volumes of traffic over inter-datacenter WAN. The Flow Completion Tim... more —Long flows contribute huge volumes of traffic over inter-datacenter WAN. The Flow Completion Time (FCT) is a vital network performance metric that affects the running time of distributed applications and the users' quality of experience. Flow routing techniques based on propagation or queuing latency or instantaneous link utilization are insufficient for minimization of the long flows' FCT. We propose a routing approach that uses the remaining sizes and paths of all ongoing flows to minimize the worst-case completion time of incoming flows assuming no knowledge of future flow arrivals. Our approach can be formulated as an NP-Hard graph optimization problem. We propose BWRH, a heuristic to quickly generate an approximate solution. We evaluate BWRH against several real WAN topologies and two different traffic patterns. We see that BWRH provides solutions with an average optimality gap of less than 0.25%. Furthermore, we show that compared to other popular routing heuristics, BWRH reduces the mean and tail FCT by up to 1.46× and 1.53×, respectively.
—Inter-datacenter networks connect dozens of geographically dispersed datacenters and carry traff... more —Inter-datacenter networks connect dozens of geographically dispersed datacenters and carry traffic flows with highly variable sizes and different classes. Adaptive flow routing can improve efficiency and performance by assigning paths to new flows according to network status and flow properties. A popular approach widely used for traffic engineering is based on current bandwidth utilization of links. We propose an alternative that reduces bandwidth usage by up to at least 50% and flow completion times by up to at least 40% across various scheduling policies and flow size distributions.
—Large inter-datacenter transfers are crucial for cloud service efficiency and are increasingly u... more —Large inter-datacenter transfers are crucial for cloud service efficiency and are increasingly used by organizations that have dedicated wide area networks between datacen-ters. A recent work uses multicast forwarding trees to reduce the bandwidth needs and improve completion times of point-to-multipoint transfers. Using a single forwarding tree per transfer, however, leads to poor performance because the slowest receiver dictates the completion time for all receivers. Using multiple forwarding trees per transfer alleviates this concern–the average receiver could finish early; however, if done naively, bandwidth usage would also increase and it is apriori unclear how best to partition receivers, how to construct the multiple trees and how to determine the rate and schedule of flows on these trees. This paper presents QuickCast, a first solution to these problems. Using simulations on real-world network topologies, we see that QuickCast can speed up the average receiver's completion time by as much as 10× while only using 1.04× more bandwidth; further, the completion time for all receivers also improves by as much as 1.6× faster at high loads.
IEEE Communications Surveys & Tutorials, May 24, 2018
—Datacenters provide cost-effective and flexible access to scalable compute and storage resources... more —Datacenters provide cost-effective and flexible access to scalable compute and storage resources necessary for today's cloud computing needs. A typical datacenter is made up of thousands of servers connected with a large network and usually managed by one operator. To provide quality access to the variety of applications and services hosted on datacenters and maximize performance, it deems necessary to use datacenter networks effectively and efficiently. Datacenter traffic is often a mix of several classes with different priorities and requirements. This includes user-generated interactive traffic, traffic with deadlines, and long-running traffic. To this end, custom transport protocols and traffic management techniques have been developed to improve datacenter network performance. In this tutorial paper, we review the general architecture of datacenter networks, various topologies proposed for them, their traffic properties, general traffic control challenges in datacenters and general traffic control objectives. The purpose of this paper is to bring out the important characteristics of traffic control in datacenters and not to survey all existing solutions (as it is virtually impossible due to massive body of existing research). We hope to provide readers with a wide range of options and factors while considering a variety of traffic control mechanisms. We discuss various characteristics of datacenter traffic control including management schemes, transmission control, traffic shaping, prioritization, load balancing, multipathing, and traffic scheduling. Next, we point to several open challenges as well as new and interesting networking paradigms. At the end of this paper, we briefly review inter-datacenter networks that connect geographically dispersed datacenters which have been receiving increasing attention recently and pose interesting and novel research problems. To measure the performance of datacenter networks, different performance metrics have been used such as flow completion times, deadline miss rate, throughput and fairness. Depending on the application and user requirements, some metrics may need more attention. While investigating different traffic control techniques, we point out the trade-offs involved in terms of costs, complexity and performance. We find that a combination of different traffic control techniques may be necessary at particular entities and layers in the network to improve the variety of performance metrics. We also find that despite significant research efforts, there are still open problems that demand further attention from the research community.
9th USENIX Workshop on Hot Topics in Cloud Computing, Jul 12, 2017
Using multiple datacenters allows for higher availability, load balancing and reduced latency to ... more Using multiple datacenters allows for higher availability, load balancing and reduced latency to customers of cloud services. To distribute multiple copies of data, cloud providers depend on inter-datacenter WANs that ought to be used efficiently considering their limited capacity and the ever-increasing data demands. In this paper, we focus on applications that transfer objects from one datacenter to several datacenters over dedicated inter-datacenter networks. We present DCCast, a centralized Point to Multi-Point (P2MP) algorithm that uses forwarding trees to efficiently deliver an object from a source datacenter to required destination datacenters. With low computational overhead, DCCast selects forwarding trees that minimize bandwidth usage and balance load across all links. With simulation experiments on Google’s GScale network, we show that DCCast can reduce total bandwidth usage and tail Transfer Completion Times (TCT) by up to 50% compared to delivering the same objects via independent point-to-point (P2P) transfers.
Networking and Internet Architecture (cs.NI), arXiv:1312.2241, Dec 8, 2013
Agent-Based Modeling and Simulation (ABMS) is a simple and yet powerful method for simulation of ... more Agent-Based Modeling and Simulation (ABMS) is a simple and yet powerful method for simulation of interactions among individual agents. Using ABMS, different phenomena can be modeled and simulated without spending additional time on unnecessary complexities. Although ABMS is well-matured in many different fields such as economic, social, and natural phenomena, it has not received much attention in the context of mobile ad-hoc networks (MANETs). In this paper, we present ABMQ, a powerful Agent-Based platform suitable for modeling and simulation of self-organization in wireless networks, and particularly MANETs. By utilizing the unique potentials of Qt Application Framework, ABMQ provides the ability to easily model and simulate self-organizing algorithms, and then reuse the codes and models developed during simulation process for building real third-party applications for several desktop and mobile platforms, which substantially decreases the development time and cost, and prevents probable bugs that can happen as a result of rewriting codes.
2013 IFIP Wireless Days (WD), 2013
Task-based self-organizing algorithms have been proposed as a solution for the management of mobi... more Task-based self-organizing algorithms have been proposed as a solution for the management of mobile ad-hoc networks (MANETs). Such algorithms assign tasks to nodes by sequentially selecting the best nodes as leaders. Since the correct functionality of such algorithms depends on the truthfulness of participant nodes, any misbehavior can disrupt the operation of the network due to absence of a centralized monitoring unit. In this paper, we consider the problem of malicious behavior in leader-based MANETs. Current solution analyzes the behavior of the leader by means of some checker nodes. This approach is vulnerable since a malicious checker can ruin the character of a normal behaving leader, and declare it as a malicious one. We propose an efficient self-organizing mechanism which can detect a malicious behaving leader, while protecting a normal behaving leader from being declared as a malicious node. In addition, our mechanism is designed with no constraints on leader selection algorithm, which makes it applicable to any kind of leader-based network. We also provide several simulation results, showing that our proposed mechanism is efficient and effective even for large MANETs.
2013 IFIP Wireless Days (WD), 2013
In this paper, we study the problem of leader selection in the presence of selfish nodes in mobil... more In this paper, we study the problem of leader selection in the presence of selfish nodes in mobile ad-hoc networks (MANETs). In order to encourage selfish nodes to assume the leadership role, some form of incentive mechanism is required. We present an incentive-based leader selection mechanism in which incentives are incorporated in the form of credit transfer to the leader, which motivates nodes to compete with each other for assuming the leadership role. The competition among nodes is modeled as a series of one-on-one incomplete information, alternating offers bargaining games. Furthermore, we propose an efficient algorithm in order to reduce the communication overhead imposed on the network for selecting a new leader, in case the current leader is disconnected from the network due to reasons such as mobility and battery depletion. Simulation results show that the proposed mechanism not only increases the overall lifetime of a network, but also decreases the amount of imposed communication overhead on the network in comparison with the traditional leader selection algorithms.
World Automation Congress (WAC), Oct 6, 2016
Datacenter-based Cloud Computing services provide a flexible, scalable and yet economical infrast... more Datacenter-based Cloud Computing services provide a flexible, scalable and yet economical infrastructure to host online services such as multimedia streaming, email and bulk storage. Many such services perform geo-replication to provide necessary quality of service and reliability to users resulting in frequent large inter-datacenter transfers. In order to meet tenant service level agreements (SLAs), these transfers have to be completed prior to a deadline. In addition, WAN resources are quite scarce and costly, meaning they should be fully utilized. Several recently proposed schemes, such as B4 [1], TEMPUS [2], and SWAN [3] have focused on improving the utilization of inter-datacenter transfers through centralized scheduling, however, they fail to provide a mechanism to guarantee that admitted requests meet their deadlines. Also, in a recent study, authors propose Amoeba [4], a system that allows tenants to define deadlines and guarantees that the specified deadlines are met, however, to admit new traffic, the proposed system has to modify the allocation of already admitted transfers. In this paper, we propose Rapid Close to Deadline Scheduling (RCD), a close to deadline traffic allocation technique that is fast and efficient. Through simulations, we show that RCD is up to 15 times faster than Amoeba, provides high link utilization along with deadline guarantees, and is able to make quick decisions on whether a new request can be fully satisfied before its deadline.
High Performance Computing (HiPC), 2016 IEEE 23rd International Conference on, Feb 2, 2017
Datacenters provide the infrastructure for cloud computing services used by millions of users eve... more Datacenters provide the infrastructure for cloud computing services used by millions of users everyday. Many such services are distributed over multiple datacenters at geographically distant locations possibly in different continents. These datacenters are then connected through high speed WAN links over private or public networks. To perform data backups or data synchronization operations, many transfers take place over these networks that have to be completed before a deadline in order to provide necessary service guarantees to end users. Upon arrival of a transfer request, we would like the system to be able to decide whether such a request can be guaranteed successful delivery. If yes, it should provide us with transmission schedule in the shortest time possible. In addition, we would like to avoid packet reordering at the destination as it affects TCP performance. Previous work in this area either cannot guarantee that admitted transfers actually finish before the specified deadlines or use techniques that can result in packet reordering. In this paper, we propose DCRoute, a fast and efficient routing and traffic allocation technique that guarantees transfer completion before deadlines for admitted requests. It assigns each transfer a single path to avoid packet reordering. Through simulations, we show that DCRoute is at least 200 times faster than other traffic allocation techniques based on linear programming (LP) while admitting almost the same amount of traffic to the system.
Conference Presentations by Mohammad Noormohammadpour
HPDC PhD Forum, 2019
This poster presents the basic idea behind the Best Worst-case Routing (BWR) which allows the ope... more This poster presents the basic idea behind the Best Worst-case Routing (BWR) which allows the operators to assign paths to new flows according to the remaining volumes of other ongoing flows. The BWR problem is NP-Hard. We present two heuristics of BWRH and BWRHF that use an upper bound on the worst-case completion time of flows on potential paths to find a path with the best worst-case completion time for a new flow. We first show that BWRHF and BWRH offer almost identical gains in the tail and mean flow completion times. Compared to two link utilization based routing techniques, BWRHF can reduce the tail completion times of flows by over 2×, and the mean completion times of flows by over 1.25×, while using up to about 50% less total network capacity.
As applications become more distributed to improve user experience and offer higher availability,... more As applications become more distributed to improve user experience and offer higher availability, businesses rely on geographically dispersed datacenters that host such applications more than ever. Dedicated inter-datacenter (inter-DC) networks have been built that provide high visibility into the network status and flexible control over traffic forwarding to offer quality communication across the instances of applications hosted on many datacenters. Using coordinated data transmission from the services and routing over the inter-DC network, one can optimize the network performance according to a variety of utility functions that take into account data transfer deadlines, network capacity consumption, and transfer completion times. Such optimization is especially relevant for bulk data transfers that occur across datacenters due to the replication of configuration data, multimedia content, and machine learning models. In this presentation, we first offer an overview of inter-DC networks and present a general framework for inter-DC traffic engineering. Next, we focus on inter-DC Point to Multipoint (P2MP) transfers, which constitute the bulk of inter-DC traffic and are created as a result of frequent data and content replication across many datacenters. We discuss multiple solutions we developed for fast and efficient P2MP transfers, namely DCCast, QuickCast, and Iris. We first show that using DCCast, which selects minimum weight Steiner trees according to current network conditions, we were able to reduce the total bandwidth usage and tail completion times of P2MP transfers by about 50% over the GScale network (Google's Inter-DC Backbone). We then present QuickCast, which aims to minimize the mean receiver completion times for P2MP transfers by separating faster and slower receivers with minimum extra bandwidth consumption. QuickCast allows the faster 50% of P2MP receivers to complete between 3× to 35× faster on average compared with when a single forwarding multicast tree is used for data delivery. We will also discuss how Iris can further improve on QuickCast and how we can further speed up the receivers using parallel forwarding trees.
Large inter-datacenter transfers are crucial for cloud service efficiency and are increasingly us... more Large inter-datacenter transfers are crucial for cloud service efficiency and are increasingly used by organizations that have dedicated wide area networks between datacenters. A recent work uses multicast forwarding trees to reduce the bandwidth needs and improve completion times of point-to-multipoint transfers. Using a single forwarding tree per transfer, however, leads to poor performance because the slowest receiver dictates the completion time for all receivers. Using multiple forwarding trees per transfer alleviates this concern--the average receiver could finish early; however, if done naively, bandwidth usage would also increase and it is apriori unclear how best to partition receivers, how to construct the multiple trees and how to determine the rate and schedule of flows on these trees. This paper presents QuickCast, a first solution to these problems. Using simulations on real-world network topologies, we see that QuickCast can speed up the average receiver's completion time by as much as 10× while only using 1.04× more bandwidth; further, the completion time for all receivers also improves by as much as 1.6× faster at high loads.
Using multiple datacenters allows for higher availability, load balancing and reduced latency to ... more Using multiple datacenters allows for higher availability, load balancing and reduced latency to customers of cloud services. To distribute multiple copies of data, cloud providers depend on inter-datacenter WANs that ought to be used efficiently considering their limited capacity and the ever-increasing data demands. In this paper, we focus on applications that transfer objects from one datacenter to several datacenters over dedicated inter-datacenter networks. We present DCCast, a centralized Point to Multi-Point (P2MP) algorithm that uses forwarding trees to efficiently deliver an object from a source datacenter to required destination datacenters. With low computational overhead, DCCast selects forwarding trees that minimize bandwidth usage and balance load across all links. With simulation experiments on Google’s GScale network, we show that DCCast can reduce total bandwidth usage and tail Transfer Completion Times (TCT) by up to 50% compared to delivering the same objects via independent point-to-point (P2P) transfers.
Uploads
Papers by Mohammad Noormohammadpour
Conference Presentations by Mohammad Noormohammadpour