Abhishek Verma

Bio: Abhishek Verma is an academic researcher from University of Illinois at Urbana–Champaign. The author has contributed to research in topics: Scheduling (computing) & Job scheduler. The author has an hindex of 23, co-authored 41 publications receiving 3003 citations. Previous affiliations of Abhishek Verma include Tata Consultancy Services & Google.

Large-scale cluster management at Google with Borg

Abstract: Google's Borg system is a cluster manager that runs hundreds of thousands of jobs, from many thousands of different applications, across a number of clusters each with up to tens of thousands of machines. It achieves high utilization by combining admission control, efficient task-packing, over-commitment, and machine sharing with process-level performance isolation. It supports high-availability applications with runtime features that minimize fault-recovery time, and scheduling policies that reduce the probability of correlated failures. Borg simplifies life for its users by offering a declarative job specification language, name service integration, real-time job monitoring, and tools to analyze and simulate system behavior. We present a summary of the Borg system architecture and features, important design decisions, a quantitative analysis of some of its policy decisions, and a qualitative examination of lessons learned from a decade of operational experience with it.

ARIA: automatic resource inference and allocation for mapreduce environments

Abstract: MapReduce and Hadoop represent an economically compelling alternative for efficient large scale data processing and advanced analytics in the enterprise. A key challenge in shared MapReduce clusters is the ability to automatically tailor and control resource allocations to different applications for achieving their performance goals. Currently, there is no job scheduler for MapReduce environments that given a job completion deadline, could allocate the appropriate amount of resources to the job so that it meets the required Service Level Objective (SLO). In this work, we propose a framework, called ARIA, to address this problem. It comprises of three inter-related components. First, for a production job that is routinely executed on a new dataset, we build a job profile that compactly summarizes critical performance characteristics of the underlying application during the map and reduce stages. Second, we design a MapReduce performance model, that for a given job (with a known profile) and its SLO (soft deadline), estimates the amount of resources required for job completion within the deadline. Finally, we implement a novel SLO-based scheduler in Hadoop that determines job ordering and the amount of resources to allocate for meeting the job deadlines.We validate our approach using a set of realistic applications. The new scheduler effectively meets the jobs' SLOs until the job demands exceed the cluster resources. The results of the extensive simulation study are validated through detailed experiments on a 66-node Hadoop cluster.

Scaling Genetic Algorithms Using MapReduce

Abstract: Genetic algorithms(GAs) are increasingly being applied to large scale problems. The traditional MPI-based parallel GAs require detailed knowledge about machine architecture. On the other hand, MapReduce is a powerful abstraction proposed by Google for making scalable and fault tolerant applications. In this paper, we show how genetic algorithms can be modeled into the MapReduce model. We describe the algorithm design and implementation of GAs on Hadoop, an open source implementation of MapReduce. Our experiments demonstrate the convergence and scalability up to 10^5 variable problems. Adding more resources would enable us to solve even larger problems without any changes in the algorithms and implementation since we do not introduce any performance bottlenecks.

Resource provisioning framework for mapreduce jobs with performance goals

Abstract: Many companies are increasingly using MapReduce for efficient large scale data processing such as personalized advertising, spam detection, and different data mining tasks. Cloud computing offers an attractive option for businesses to rent a suitable size Hadoop cluster, consume resources as a service, and pay only for resources that were utilized. One of the open questions in such environments is the amount of resources that a user should lease from the service provider. Often, a user targets specific performance goals and the application needs to complete data processing by a certain time deadline. However, currently, the task of estimating required resources to meet application performance goals is solely the users' responsibility. In this work, we introduce a novel framework and technique to address this problem and to offer a new resource sizing and provisioning service in MapReduce environments. For a MapReduce job that needs to be completed within a certain time, the job profile is built from the job past executions or by executing the application on a smaller data set using an automated profiling tool. Then, by applying scaling rules combined with a fast and efficient capacity planning model, we generate a set of resource provisioning options. Moreover, we design a model for estimating the impact of node failures on a job completion time to evaluate worst case scenarios. We validate the accuracy of our models using a set of realistic applications. The predicted completion times of generated resource provisioning options are within 10% of the measured times in our 66-node Hadoop cluster.

Resource provisioning framework for mapreduce jobs with performance goals

Abstract: Many companies are increasingly using MapReduce for efficient large scale data processing such as personalized advertising, spam detection, and different data mining tasks. Cloud computing offers an attractive option for businesses to rent a suitable size Hadoop cluster, consume resources as a service, and pay only for resources that were utilized. One of the open questions in such environments is the amount of resources that a user should lease from the service provider. Often, a user targets specific performance goals and the application needs to complete data processing by a certain time deadline. However, currently, the task of estimating required resources to meet application performance goals is solely the users' responsibility. In this work, we introduce a novel framework and technique to address this problem and to offer a new resource sizing and provisioning service in MapReduce environments. For a MapReduce job that needs to be completed within a certain time, the job profile is built from the job past executions or by executing the application on a smaller data set using an automated profiling tool. Then, by applying scaling rules combined with a fast and efficient capacity planning model, we generate a set of resource provisioning options. Moreover, we design a model for estimating the impact of node failures on a job completion time to evaluate worst case scenarios. We validate the accuracy of our models using a set of realistic applications. The predicted completion times of generated resource provisioning options are within 10% of the measured times in our 66-node Hadoop cluster.

TensorFlow: a system for large-scale machine learning

Abstract: TensorFlow is a machine learning system that operates at large scale and in heterogeneous environments. Tensor-Flow uses dataflow graphs to represent computation, shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general-purpose GPUs, and custom-designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer: whereas in previous "parameter server" designs the management of shared state is built into the system, TensorFlow enables developers to experiment with novel optimizations and training algorithms. TensorFlow supports a variety of applications, with a focus on training and inference on deep neural networks. Several Google services use TensorFlow in production, we have released it as an open-source project, and it has become widely used for machine learning research. In this paper, we describe the TensorFlow dataflow model and demonstrate the compelling performance that TensorFlow achieves for several real-world applications.

TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

Abstract: TensorFlow is an interface for expressing machine learning algorithms, and an implementation for executing such algorithms. A computation expressed using TensorFlow can be executed with little or no change on a wide variety of heterogeneous systems, ranging from mobile devices such as phones and tablets up to large-scale distributed systems of hundreds of machines and thousands of computational devices such as GPU cards. The system is flexible and can be used to express a wide variety of algorithms, including training and inference algorithms for deep neural network models, and it has been used for conducting research and for deploying machine learning systems into production across more than a dozen areas of computer science and other fields, including speech recognition, computer vision, robotics, information retrieval, natural language processing, geographic information extraction, and computational drug discovery. This paper describes the TensorFlow interface and an implementation of that interface that we have built at Google. The TensorFlow API and a reference implementation were released as an open-source package under the Apache 2.0 license in November, 2015 and are available at www.tensorflow.org.

TensorFlow: A system for large-scale machine learning

Abstract: TensorFlow is a machine learning system that operates at large scale and in heterogeneous environments. TensorFlow uses dataflow graphs to represent computation, shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general-purpose GPUs, and custom designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer: whereas in previous "parameter server" designs the management of shared state is built into the system, TensorFlow enables developers to experiment with novel optimizations and training algorithms. TensorFlow supports a variety of applications, with particularly strong support for training and inference on deep neural networks. Several Google services use TensorFlow in production, we have released it as an open-source project, and it has become widely used for machine learning research. In this paper, we describe the TensorFlow dataflow model in contrast to existing systems, and demonstrate the compelling performance that TensorFlow achieves for several real-world applications.

Network Function Virtualization: State-of-the-Art and Research Challenges

Abstract: Network function virtualization (NFV) has drawn significant attention from both industry and academia as an important shift in telecommunication service provisioning. By decoupling network functions (NFs) from the physical devices on which they run, NFV has the potential to lead to significant reductions in operating expenses (OPEX) and capital expenses (CAPEX) and facilitate the deployment of new services with increased agility and faster time-to-value. The NFV paradigm is still in its infancy and there is a large spectrum of opportunities for the research community to develop new architectures, systems and applications, and to evaluate alternatives and trade-offs in developing technologies for its successful deployment. In this paper, after discussing NFV and its relationship with complementary fields of software defined networking (SDN) and cloud computing, we survey the state-of-the-art in NFV, and identify promising research directions in this area. We also overview key NFV projects, standardization efforts, early implementations, use cases, and commercial products.

Large-scale cluster management at Google with Borg

Abstract: Google's Borg system is a cluster manager that runs hundreds of thousands of jobs, from many thousands of different applications, across a number of clusters each with up to tens of thousands of machines. It achieves high utilization by combining admission control, efficient task-packing, over-commitment, and machine sharing with process-level performance isolation. It supports high-availability applications with runtime features that minimize fault-recovery time, and scheduling policies that reduce the probability of correlated failures. Borg simplifies life for its users by offering a declarative job specification language, name service integration, real-time job monitoring, and tools to analyze and simulate system behavior. We present a summary of the Borg system architecture and features, important design decisions, a quantitative analysis of some of its policy decisions, and a qualitative examination of lessons learned from a decade of operational experience with it.