Shivaram Venkataraman is a PhD Candidate at the University of California, Berkeley and works with Mike Franklin and Ion Stoica. He is a committer on the Apache Spark project and his research interests are in designing systems for large scale machine-learning. Before coming to Berkeley, he completed his M.S at the University of Illinois, Urbana-Champaign and worked as a Software Engineer at Google.
The BigDL framework scales deep learning for large data sets using Apache Spark. However there is significant scheduling overhead from Spark when running BigDL at large scale. In this talk we propose a new parameter manager implementation that along with coarse-grained scheduling can provide significant speedups for deep learning models like Inception, VGG etc. Aggregation functions like reduce or treeReduce that are used for parameter aggregation in Apache Spark (and the original MapReduce) are slow as the centralized scheduling and driver network bandwidth become a bottleneck especially in large clusters. To reduce the overhead of parameter aggregation and allow for near-linear scaling, we introduce a new AllReduce operation, a part of the parameter manager in BigDL which is built directly on top of the BlockManager in Apache Spark. AllReduce in BigDL uses a peer-to-peer mechanism to synchronize and aggregate parameters. During parameter synchronization and aggregation, all nodes in the cluster play the same role and driver's overhead is eliminated thus enabling near-linear scaling. To address the scheduling overhead we use Drizzle, a recently proposed scheduling framework for Apache Spark. Currently, Spark uses a BSP computation model, and notifies the scheduler at the end of each task. Invoking the scheduler at the end of each task adds overheads and results in decreased throughput and increased latency. Drizzle introduces group scheduling, where multiple iterations (or a group) of iterations are scheduled at once. This helps decouple the granularity of task execution from scheduling and amortizes the costs of task serialization and launch. Finally we will present results from using the new AllReduce operation and Drizzle on a number of common deep learning models including VGG and Inception. Our benchmarks run on Amazon EC2 and Google DataProc will show the speedups and scalability of our implementation.
R is a widely used statistical programming language but its interactive use is typically limited to a single machine. To enable large scale data analysis from R, we will present SparkR, an open source R package developed at UC Berkeley, that allows data scientists to analyze large data sets and interactively run jobs on them from the R shell. This talk will introduce SparkR, discuss some of its features and highlight the power of combining R’s interactive console and extension packages with Spark’s distributed run-time.