Recently, network coding has become a promising transmission approach to support high throughput and low latency in distributed multihop networks. In this article, we develop a delay-controlled distributed route establishment scheme that can provide maximal bidirectional transmission to enhance network coding gain while satisfying a time-critical route setup. The scheme is called network coding-aware delayed store and forwarding (NC-DSF). It delays the received route information packets before forwarding them according to the link status and network topology. We propose a tight delay function derived using a strict end-to-end delay bound for delay control. Subsequently, we suggest a relaxed delay function derived using realistic and practical conditions. Finally, we propose a load-weighted delay function considering the tradeoff between bidirectionality and network-load balancing. The simulations confirm that the proposed scheme offers increased throughput and decreased latency in mesh and random multihop networks. The proposed transmission scheme, NC-DSF, can be efficiently employed in the future industrial Internet of Things networks requiring a time-constrained route setup, high throughput, and low latency.

We introduce Harmony, a new scheduling framework that executes multiple Parameter-Server ML training jobs together to improve cluster resource utilization. Harmony coordinates a fine-grained execution of co-located jobs with complementary resource usages to avoid contention and to efficiently share resources between the jobs. To resolve the memory pressure due to the increased number of simultaneous jobs, Harmony uses a data spill/reload mechanism optimized for multiple jobs with the iterative execution pattern. Our evaluation shows that Harmony improves cluster resource utilization by up to 1.65×, resulting in a reduction of the mean ML training job time by about 53%, and makespan, the total time to process all given jobs, by about 38%, compared to the traditional approaches that allocate dedicated resources to each job.

The performance of distributed machine learning systems is dependent on their system configuration. However, configuring the system for optimal performance is challenging and time consuming even for experts due to the diverse runtime factors such as workloads or the system environment. We present cost-based optimization to automatically find a good system configuration for parameter server (PS) machine learning (ML) frameworks. We design and implement Cruise that applies the optimization technique to tune distributed PS ML execution automatically. Evaluation results on three ML applications verify that Cruise automates the system configuration of the applications to achieve good performance with minor reconfiguration costs.

With the proliferation of Internet of Things (IoT) devices, the load caused by network demands has been increasing. This has direct consequences in terms of increased network requests and thus significantly higher energy consumption. To this end, this paper focuses on network power saving. In a cloud-radio access network with fiber-wireless, we propose a power saving method based on the optical line terminals (OLTs) and optical network units (ONUs), the equipment responsible for data transmission to and from the devices. Our proposed method is based on the novel idea of dividing the sleep phase of the ONUs into time slots, which facilitates synchronization and coordination in the whole system, thus increasing resource efficiency and bringing power economy not only to the ONUs, but also to the OLTs and the network as a whole. This is in contrast to the existing literature, which considers only ONUs or OLTs separately, without focusing on the cooperation within the system. Additionally, our proposed method was created with IoT traffic in mind. and it was designed to guarantee a maximum delay in order to make it applicable to mobile and real-time IoT. Finally, our proposed method allows the network to operate without external power sources for the ONUs by using power over fiber in conjunction with the OLT. Mathematical analyses show that our proposed method can successfully decrease the system’s energy consumption.

Internet of Things (IoT) technology has accelerated various industries through digital transformation. In an edge computing-based smart factory, a significant number of IoT devices generate large volumes of real-time data. This big data requires efficient routing among edge gateways (EGs) and an edge server for real-time data processing. Existing industrial wireless communication systems provide relatively low data rates and network capacity for real-time sensor data and control information over a wireless channel. This calls for the use of the very large bandwidth available at the mmWave spectrum for real-time data transmission. Existing data routing techniques for the mmWave band are based on traditional mobile ad hoc routing techniques and do not reduce the transmission delay for real-time sensory data in smart manufacturing systems. Therefore, to alleviate the real-time data processing requirement, we propose a new directional routing and link scheduling algorithm based on maximum weight independent set (MWIS). The proposed algorithm solves complicated MWIS problems efficiently and computes backhaul link scheduling results in a relatively short time by lowering the deafness problem among EGs. For transmission fairness, we used a Jain's fairness index method with numerical analysis of the transmission fairness constraint. We measured the efficiency of our proposed scheme in terms of throughput, delay, packet loss rate, and transmission fairness. Our simulation results show that the proposed scheme outperforms existing mmWave routing techniques. Moreover, we investigated the performance difference between the proposed algorithm and the optimal solution.

Datacenters are under-utilized, primarily due to unused resources on over-provisioned nodes of latency-critical jobs. Such idle resources can be used to run batch data analytic jobs to increase datacenter utilization, but these transient resources must be evicted whenever latency-critical jobs require them again. Resource evictions often lead to cascading recomputations, which is usually handled by checkpointing intermediate results on stable storages of eviction-free reserved resources. However, checkpointing has major shortcomings in its substantial overhead of transferring data back and forth. In this work, we step away from such approaches and focus on observing the job structure and the relationships between computations of the job. We carefully mark the computations that are most likely to cause a large number of recomputations upon evictions, to run them reliably using reserved resources. This lets us retain corresponding intermediate results effortlessly without any additional checkpointing. We design Pado, a general data processing engine, which carries out our idea with several optimizations that minimize the number of additional reserved nodes. Evaluation results show that Pado outperforms Spark 2.0.0 by up to 5.1×, and checkpoint-enabled Spark by up to 3.8×.

Machine Learning models are often composed by sequences of transformations. While this design makes easy to decompose and accelerate single model components at training time, predictions requires low latency and high performance predictability whereby end-to-end runtime optimizations and acceleration is needed to meet such goals. This paper shed some light on the problem by using a production-like model, and showing how by redesigning model pipelines for efficient execution over CPUs and FPGAs performance improvements of several folds can be achieved.

Resource Managers like YARN and Mesos have emerged as a critical layer in the cloud computing system stack, but the developer abstractions for leasing cluster resources and instantiating application logic are very low level. This flexibility comes at a high cost in terms of developer effort, as each application must repeatedly tackle the same challenges (e.g., fault tolerance, task scheduling and coordination) and reimplement common mechanisms (e.g., caching, bulk-data transfers). This article presents REEF, a development framework that provides a control plane for scheduling and coordinating task-level (data-plane) work on cluster resources obtained from a Resource Manager. REEF provides mechanisms that facilitate resource reuse for data caching and state management abstractions that greatly ease the development of elastic data processing pipelines on cloud platforms that support a Resource Manager service. We illustrate the power of REEF by showing applications built atop: a distributed shell application, a machine-learning framework, a distributed in-memory caching system, and a port of the CORFU system. REEF is currently an Apache top-level project that has attracted contributors from several institutions and it is being used to develop several commercial offerings such as the Azure Stream Analytics service.

Resource Managers like Apache YARN have emerged as a critical layer in the cloud computing system stack, but the developer abstractions for leasing cluster resources and instantiating application logic are very low-level. This flexibility comes at a high cost in terms of developer effort, as each application must repeatedly tackle the same challenges (e.g., fault-tolerance, task scheduling and coordination) and re-implement common mechanisms (e.g., caching, bulk-data transfers). This paper presents REEF, a development framework that provides a control-plane for scheduling and coordinating task-level (data-plane) work on cluster resources obtained from a Resource Manager. REEF provides mechanisms that facilitate resource re-use for data caching, and state management abstractions that greatly ease the development of elastic data processing work-flows on cloud platforms that support a Resource Manager service. REEF is being used to develop several commercial offerings such as the Azure Stream Analytics service. Furthermore, we demonstrate REEF development of a distributed shell application, a machine learning algorithm, and a port of the CORFU [4] system. REEF is also currently an Apache Incubator project that has attracted contributors from several instititutions.