arXiv subjects

Suresh Subramaniam

Publications and source records attributed to Suresh Subramaniam.

28 records · Page 2Linked to original sources

arXiv2020

Maintaining Information Freshness in Power-Efficient Status Update Systems

This paper is motivated by emerging edge computing systems which consist of sensor nodes that acquire and process information and then transmit status updates to an edge receiver for possible further processing. As power is a scarce resource at the sensor nodes, the system is modeled as a tandem computation-transmission queue with power-efficient computing. Jobs arrive at the computation server with rate $λ$ as a Poisson process with no available data buffer. The computation server can be in one of three states: (i) OFF: the server is turned off and no jobs are observed or processed, (ii) ON-Idle: the server is turned on but there is no job in the server, (iii) ON-Busy: the server is turned on and a job is processed in the server. These states cost zero, one and $p_c$ units of power, respectively. Under a long-term power constraint, the computation server switches from one state to another in sequence: first a deterministic $T_o$ time units in OFF state, then waiting for a job arrival in ON-Idle state and then in ON-Busy state for an independent identically distributed compute time duration. The transmission server has a single unit data buffer to save incoming packets and applies last come first serve with discarding as well as a packet deadline to discard a sitting packet for maintaining information freshness, which is measured by the Age of Information (AoI). Additionally, there is a monotonic functional relation between the mean time spent in ON-Busy state and the mean transmission time. We obtain closed-form expressions for average AoI and average peak AoI. Our numerical results illustrate various regimes of operation for best AoI performances optimized over packet deadlines with relation to power efficiency.

cs.IT↗

arXiv2019

Optimizing Information Freshness Through Computation-Transmission Tradeoff and Queue Management in Edge Computing

Edge computing applications typically require generated data to be preprocessed at the source and then transmitted to an edge server. In such cases, transmission time and preprocessing time are coupled, yielding a tradeoff between them to achieve the targeted objective. This paper presents analysis of such a system with the objective of optimizing freshness of received data at the edge server. We model this system as two queues in tandem whose service times are independent over time but the transmission service time is monotonically dependent on the computation service time in mean value. This dependence captures the natural decrease in transmission time due to lower offloaded computation. We analyze various queue management schemes in this tandem queue where the first queue has a single server, Poisson packet arrivals, general independent service and no extra buffer to save incoming status update packets. The second queue has a single server receiving packets from the first queue and service is memoryless. We consider the second queue in two forms: (i) No data buffer and (ii) One unit data buffer and last come first serve with discarding. We analyze various non-preemptive as well as preemptive cases. We perform stationary distribution analysis and obtain closed form expressions for average age of information (AoI) and average peak AoI. Our numerical results illustrate analytical findings on how computation and transmission times could be traded off to optimize AoI and reveal a consequent tradeoff between average AoI and average peak AoI.

cs.IT↗

arXiv2019

HolDCSim: A Holistic Simulator for Data Centers

Cloud computing based systems, that span data centers, are commonly deployed to offer high performance for user service requests. As data centers continue to expand, computer architects and system designers are facing many challenges on how to balance resource utilization efficiency, server and network performance, energy consumption and quality-of-service (QoS) demands from the users. To develop effective data center management policies, it becomes essential to have an in-depth understanding and synergistic control of the various sub-components inside large scale computing systems, that include both computation and communication resources. In this paper, we propose HolDCSim, a light-weight, holistic, extensible, event-driven data center simulation platform that effectively models both server and network architectures. HolDCSim can be used in a variety of data center system studies including job/task scheduling, resource provisioning, global and local server farm power management, and network and server performance analysis. We demonstrate the design of our simulation infrastructure, and illustrate the usefulness of our framework with several case studies that analyze server/network performance and energy efficiency. We also perform validation on real machines to verify our simulator.

cs.DC↗

arXiv2019

Relative Age of Information: A New Metric for Status Update Systems

In this paper, we introduce a new data freshness metric, relative Age of Information (rAoI), and examine it in a single server system with various packet management schemes. The (classical) AoI metric was introduced to measure the staleness of status updates at the receiving end with respect to their generation at the source. This metric addresses systems where the timings of update generation at the source are absolute and can be designed separately or jointly with the transmission schedules. In many decentralized applications, transmission schedules are blind to update generation timing, and the transmitter can know the timing of an update packet only after it arrives. As such, an update becomes stale after a new one arrives. The rAoI metric measures how fresh the data is at the receiver with respect to the data at the transmitter. It introduces a particularly explicit dependence on the arrival process in the evaluation of age. We investigate several queuing disciplines and provide closed form expressions for rAoI and numerical comparisons.

cs.IT↗

arXiv2019

Trading Off Computation with Transmission in Status Update Systems

This paper is motivated by emerging edge computing applications in which generated data are pre-processed at the source and then transmitted to an edge server. In such a scenario, there is typically a tradeoff between the amount of pre-processing and the amount of data to be transmitted. We model such a system by considering two non-preemptive queues in tandem whose service times are independent over time but the transmission service time is dependent on the computation service time in mean value. The first queue is in M/GI/1/1 form with a single server, memoryless exponential arrivals, general independent service and no extra buffer to save incoming status update packets. The second queue is in GI/M/1/2* form with a single server receiving packets from the first queue, memoryless service and a single data buffer to save incoming packets. Additionally, mean service times of the first and second queues are dependent through a deterministic monotonic function. We perform stationary distribution analysis in this system and obtain closed form expressions for average age of information (AoI) and average peak AoI. Our numerical results illustrate the analytical findings and highlight the tradeoff between average AoI and average peak AoI generated by the tandem nature of the queueing system with dependent service times.

cs.IT↗

arXiv2019

Waiting before Serving: A Companion to Packet Management in Status Update Systems

In this paper, we explore the potential of server waiting before packet transmission in improving the Age of Information (AoI) in status update systems. We consider a non-preemptive queue with Poisson arrivals and independent general service distribution and we incorporate waiting before serving in two packet management schemes: M/GI/1/1 and M/GI/1/$2^*$. In M/GI/1/1 scheme, the server waits for a deterministic time immediately after a packet enters the server. In M/GI/1/$2^*$ scheme, depending on idle or busy system state, the server waits for a deterministic time before starting service of the packet. In both cases, if a potential newer arrival is captured existing packet is discarded. Different from most existing works, we analyze AoI evolution by indexing the incoming packets, which is enabled by an alternative method of partitioning the area under the evolution of instantaneous AoI to calculate its time average. We obtain expressions for average and average peak AoI for both queueing disciplines with waiting. Our numerical results demonstrate that waiting before service can bring significant improvement in average age, particularly, for heavy-tailed service distributions. This improvement comes at the expense of an increase in average peak AoI. We highlight the trade-off between average and average peak AoI generated by waiting before serving.

cs.IT↗

arXiv2019

A Hierarchical WDM-based Scalable Data Center Network Architecture

Massive data centers are at the heart of the Internet. The rapid growth of Internet traffic and the abundance of rich data-driven applications have raised the need for enormous network bandwidth. Towards meeting this growing traffic demand, optical interconnects have gained significant attention, as they can provide high throughput, low latency, and scalability. In particular, optical Wavelength Division Multiplexing (WDM) provides the possibility to build data centers comprising of millions of servers, while providing hundreds of terabits per second bandwidth. In this paper, we propose a WDM-based Reconfigurable Hierarchical Optical Data Center Architecture (RHODA) that can satisfy future Internet traffic demands. To improve scalability, our DCN architecture is hierarchical, as it groups server racks into clusters. Cluster membership is reconfigurable through the use of optical switches. Each cluster enables heavy-traffic communication among the racks within. To support varying traffic patterns, the inter-cluster network topology and link capacities are also reconfigurable, which is achieved through the use of optical space switches and Wavelength Selective Switches (WSSs). Our simulation results demonstrate that in terms of average hop distance, RHODA outperforms OSA, FatTree and WaveCube by up to 81%, 66% and 60%, respectively.

cs.NI↗

arXiv2019

A Reconfigurable High-Performance Optical Data Center Architecture

Optical data center network architectures are becoming attractive because of their low energy consumption, large bandwidth, and low cabling complexity. In\cite{Xu1605:PODCA}, an AWGR-based passive optical data center architecture (PODCA) is presented. Compared with other optical data center architectures, e.g., DOS \cite{ye2010scalable}, Proteus \cite{singla2010proteus}, and Petabit \cite{xia2010petabit}, PODCA can save up to 90$\%$ on power consumption and 88$\%$ in cost. Also, average latency can be low as 9 $μ$s at close to 100$\%$ throughput. However, PODCA is not reconfigurable and cannot optimize the network topology to dynamic traffic. In this paper, we present a novel, scalable and flexible reconfigurable architecture called RODCA. RODCA is built on and augments PODCA with a flexible localized intra-cluster optical network. With the reconfigurable intra-cluster network, racks with mutually large traffic can be located within the same cluster, and share the large bandwidth of the intra-cluster network. We present an algorithm for DCN topology reconfiguration, and present simulation results to demonstrate the effectiveness of reconfiguration.

cs.NI↗

arXiv2018

Chronos: A Unifying Optimization Framework for Speculative Execution of Deadline-critical MapReduce Jobs

Meeting desired application deadlines in cloud processing systems such as MapReduce is crucial as the nature of cloud applications is becoming increasingly mission-critical and deadline-sensitive. It has been shown that the execution times of MapReduce jobs are often adversely impacted by a few slow tasks, known as stragglers, which result in high latency and deadline violations. While a number of strategies have been developed in existing work to mitigate stragglers by launching speculative or clone task attempts, none of them provides a quantitative framework that optimizes the speculative execution for offering guaranteed Service Level Agreements (SLAs) to meet application deadlines. In this paper, we bring several speculative scheduling strategies together under a unifying optimization framework, called Chronos, which defines a new metric, Probability of Completion before Deadlines (PoCD), to measure the probability that MapReduce jobs meet their desired deadlines. We systematically analyze PoCD for popular strategies including Clone, Speculative-Restart, and Speculative-Resume, and quantify their PoCD in closed-form. The result illuminates an important tradeoff between PoCD and the cost of speculative execution, measured by the total (virtual) machine time required under different strategies. We propose an optimization problem to jointly optimize PoCD and execution cost in different strategies, and develop an algorithmic solution that is guaranteed to be optimal. Chronos is prototyped on Hadoop MapReduce and evaluated against three baseline strategies using both experiments and trace-driven simulations, achieving 50% net utility increase with up to 80% PoCD and 88% cost improvements.

cs.DC↗

arXiv2014

Impact of Wavelength and Modulation Conversion on Transluscent Elastic Optical Networks Using MILP

Compared to legacy wavelength division multiplexing networks, elastic optical networks (EON) have added flexibility to network deployment and management. EONs can include previously available technology, such as signal regeneration and wavelength conversion, as well as new features such as finer-granularity spectrum assignment and modulation conversion. Yet each added feature adds to the cost of the network. In order to quantify the potential benefit of each technology, we present a link-based mixed-integer linear programming (MILP) formulation to solve the optimal resource allocation problem. We then propose a recursive model in order to either augment existing network deployments or speed up the resource allocation computation time for larger networks with higher traffic demand requirements than can be solved using an MILP. We show through simulation that systems equipped with signal regenerators or wavelength converters require a notably smaller total bandwidth, depending on the topology of the network. We also show that the suboptimal recursive solution speeds up the calculation and makes the running-time more predictable, compared to the optimal MILP.

cs.NI↗