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.

/pdf/tensorflow-a-system-for-large-scale-machine-learning-10wgdjhl6k.pdf

TensorFlow: a system for large-scale machine learning

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.

/pdf/tensorflow-large-scale-machine-learning-on-heterogeneous-2yk4fgchw6.pdf

TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

As computation continues to move into the cloud, the computing platform of interest no longer resembles a pizza box or a refrigerator, but a warehouse full of computers. These new large datacenters are quite different from traditional hosting facilities of earlier times and cannot be viewed simply as a collection of co-located servers. Large portions of the hardware and software resources in these facilities must work in concert to efficiently deliver good levels of Internet service performance, something that can only be achieved by a holistic approach to their design and deployment. In other words, we must treat the datacenter itself as one massive warehouse-scale computer (WSC). We describe the architecture of WSCs, the main factors influencing their design, operation, and cost structure, and the characteristics of their software base. We hope it will be useful to architects and programmers of today's WSCs, as well as those of future many-core platforms which may one day implement the equivalent of today's WSCs on a single board. Table of Contents: Introduction / Workloads and Software Infrastructure / Hardware Building Blocks / Datacenter Basics / Energy and Power Efficiency / Modeling Costs / Dealing with Failures and Repairs / Closing Remarks

/pdf/the-datacenter-as-a-computer-an-introduction-to-the-design-nu43k2m3iz.pdf

The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines

Big data may contain big values, but also brings lots of challenges to the computing theory, architecture, framework, knowledge discovery algorithms, and domain specific tools and applications. Beyond the 4-V or 5-V characters of big datasets, the data processing shows the features like inexact, incremental, and inductive manner. This brings new research opportunities to research community across theory, systems, algorithms, and applications. Is there some new "theory" for the big data? How to handle the data computing algorithms in an operatable manner? This report shares some view on new challenges identified, and covers some of the application scenarios such as micro-blog data analysis and data processing in building next generation search engines.

/pdf/scaling-distributed-machine-learning-with-the-parameter-2g1ec41azg.pdf

Scaling Distributed Machine Learning with the Parameter Server

Modern enterprise applications are currently undergoing a complete paradigm shift away from traditional transactional processing to combined analytical and transactional processing. This challenge of combining two opposing query types in a single database management system results in additional requirements for transaction management as well. In this paper, we discuss our approach to achieve high throughput for transactional query processing while allowing concurrent analytical queries. We present our approach to distributed snapshot isolation and optimized two-phase commit protocols.

/pdf/apache-flink-stream-and-batch-processing-in-a-single-engine-3u2381y8rg.pdf

Apache flink : Stream and batch processing in a single engine

Commodity computer systems contain more and more processor cores and exhibit increasingly diverse architectural tradeoffs, including memory hierarchies, interconnects, instruction sets and variants, and IO configurations. Previous high-performance computing systems have scaled in specific cases, but the dynamic nature of modern client and server workloads, coupled with the impossibility of statically optimizing an OS for all workloads and hardware variants pose serious challenges for operating system structures.We argue that the challenge of future multicore hardware is best met by embracing the networked nature of the machine, rethinking OS architecture using ideas from distributed systems. We investigate a new OS structure, the multikernel, that treats the machine as a network of independent cores, assumes no inter-core sharing at the lowest level, and moves traditional OS functionality to a distributed system of processes that communicate via message-passing.We have implemented a multikernel OS to show that the approach is promising, and we describe how traditional scalability problems for operating systems (such as memory management) can be effectively recast using messages and can exploit insights from distributed systems and networking. An evaluation of our prototype on multicore systems shows that, even on present-day machines, the performance of a multikernel is comparable with a conventional OS, and can scale better to support future hardware.

/pdf/the-multikernel-a-new-os-architecture-for-scalable-multicore-1zaxph0v39.pdf

The multikernel: a new OS architecture for scalable multicore systems

Naiad is a distributed system for executing data parallel, cyclic dataflow programs. It offers the high throughput of batch processors, the low latency of stream processors, and the ability to perform iterative and incremental computations. Although existing systems offer some of these features, applications that require all three have relied on multiple platforms, at the expense of efficiency, maintainability, and simplicity. Naiad resolves the complexities of combining these features in one framework.A new computational model, timely dataflow, underlies Naiad and captures opportunities for parallelism across a wide class of algorithms. This model enriches dataflow computation with timestamps that represent logical points in the computation and provide the basis for an efficient, lightweight coordination mechanism.We show that many powerful high-level programming models can be built on Naiad's low-level primitives, enabling such diverse tasks as streaming data analysis, iterative machine learning, and interactive graph mining. Naiad outperforms specialized systems in their target application domains, and its unique features enable the development of new high-performance applications.

/pdf/naiad-a-timely-dataflow-system-1z2cah9a5l.pdf

Naiad: a timely dataflow system

Tools to understand complex system behaviour are essential for many performance analysis and debugging tasks, yet there are many open research problems in their development. Magpie is a toolchain for automatically extracting a system's workload under realistic operating conditions. Using low-overhead instrumentation, we monitor the system to record fine-grained events generated by kernel, middleware and application components. The Magpie request extraction tool uses an application-specific event schema to correlate these events, and hence precisely capture the control flow and resource consumption of each and every request. By removing scheduling artefacts, whilst preserving causal dependencies, we obtain canonical request descriptions from which we can construct concise workload models suitable for performance prediction and change detection. In this paper we describe and evaluate the capability of Magpie to accurately extract requests and construct representative models of system behaviour.

/pdf/using-magpie-for-request-extraction-and-workload-modelling-1gfx849p4j.pdf

Using magpie for request extraction and workload modelling

Understanding the performance of distributed systems requires correlation of thousands of interactions between numerous components -- a task best left to a computer. Today's systems provide voluminous traces from each component but do not synthesise the data into concise models of system performance.

We argue that online performance modelling should be a ubiquitous operating system service and outline several uses including performance debugging, capacity planning, system tuning and anomaly detection. We describe the Magpie modelling service which collates detailed traces from multiple machines in an e-commerce site, extracts request-specific audit trails, and constructs probabilistic models of request behaviour. A feasibility study evaluates the approach using an offline demonstrator. Results show that the approach is promising, but that there are many challenges to building a truly ubiquitious, online modelling infrastructure.

/pdf/magpie-online-modelling-and-performance-aware-systems-uwvx4bvote.pdf

Magpie: online modelling and performance-aware systems

Existing computational models for processing continuously changing input data are unable to efficiently support iterative queries except in limited special cases. This makes it difficult to perform complex tasks, such as social-graph analysis on changing data at interactive timescales, which would greatly benefit those analyzing the behavior of services like Twitter. In this paper we introduce a new model called differential computation, which extends traditional incremental computation to allow arbitrarily nested iteration, and explain—with reference to a publicly available prototype system called Naiad—how differential computation can be efficiently implemented in the context of a declarative data-parallel dataflow language. The resulting system makes it easy to program previously intractable algorithms such as incrementally updated strongly connected components, and integrate them with data transformation operations to obtain practically relevant insights from real data streams.

/pdf/differential-dataflow-5cfaay8ju5.pdf

Rebecca Isaacs

Papers

The multikernel: a new OS architecture for scalable multicore systems

Naiad: a timely dataflow system

Using magpie for request extraction and workload modelling

Magpie: online modelling and performance-aware systems

Differential dataflow