Analyzing efficient stream processing on modern hardware

Abstract

Modern Stream Processing Engines (SPEs) process large data volumes under tight latency constraints. Many SPEs execute processing pipelines using message passing on shared-nothing architectures and apply a partition-based scale-out strategy to handle high-velocity input streams. Furthermore, many state-of-the-art SPEs rely on a Java Virtual Machine to achieve platform independence and speed up system development by abstracting from the underlying hardware. In this paper, we show that taking the underlying hardware into account is essential to exploit modern hardware efficiently. To this end, we conduct an extensive experimental analysis of current SPEs and SPE design alternatives optimized for modern hardware. Our analysis highlights potential bottlenecks and reveals that state-of-the-art SPEs are not capable of fully exploiting current and emerging hardware trends, such as multi-core processors and high-speed networks. Based on our analysis, we describe a set of design changes to the common architecture of SPEs to scale-up on modern hardware. We show that the single-node throughput can be increased by up to two orders of magnitude compared to state-of-the-art SPEs by applying specialized code generation, fusing operators, batch-style parallelization strategies, and optimized windowing. This speedup allows for deploying typical streaming applications on a single or a few nodes instead of large clusters.

Hardware-Conscious Stream Processing

“…To this end, many recent works have exploited the potential of high-performance stream processing on a single node [45,57,98]. However, the important question of how best to use powerful local nodes in the context of large distributed computation setting still remains unclear.…”

“…Zeuch et al [98] analyzed the design space of DSPSs optimized for modern multicore processors. In particular, they show that a queue-less execution engine based on query compilation, which replaces communication between operators with function calls, is highly suitable for modern hardware.…”

“…Contrary to most distributed DSPSs (e.g., Storm, Flink) adopting the continuous operator model, Trill runs the whole query only on the thread that feeds data to it. Such an approach has shown to be especially effective [98] when applications contain no synchronization barriers.…”

“…Despite the success of the last several decades, more radical performance demand, complex analysis, as well as intensive state access in emerging stream applications [21,91] pose great challenges to existing DSPSs. Meanwhile, significant achievements have been made in the computer architecture community, which has recently led to various investigations of the potential of hardware-conscious DSPSs, which aim to exploit the potential of accelerating stream processing on modern hardware [104,98].…”

FogGuru: a Fog Computing platform based on Apache Flink

“…Processing: Stream Processing Engines (SPEs) represent a good paradigm for Fog computing applications because of their extensive set of features, including support for event-driven, data pipeline, and data analytics applications. Constrained resources is also here an aspect to cater for; the literature includes surveys on how various SPEs perform in a fog environments [6], [7]. We decided to use Apache Flink 2 as our SPE of choice.…”