Facebook boosts Hadoop with scheduling muscle
Facebook's Corona will make better use of clusters than MapReduce does, the company claims
By Joab Jackson | Published: 11:29, 09 November 2012
Facebook has beaten some of the limitations of the Apache Hadoop data processing platform, its engineers assert.
Facebook has released source code for scheduling workloads on the Apache Hadoop data processing platform. Engineers at the social networking company claim this program, called Corona, is superior to Hadoop's own scheduler in MapReduce.
In tests, the Corona scheduler was able to put more than 95% of a cluster to work on jobs, whereas MapReduce could utilise, at the most, 70% of a cluster, Facebook said.
Related Articles on Techworld
By using the clusters more efficiently, Facebook is able to analyse more information with existing hardware. Corona offers a number of additional benefits as well, including faster loading of workloads and a more flexible way of upgrading the software.
Facebook announced the release of Corona in a posting by a number of Facebook engineers who contributed to the software, including Avery Ching, Ravi Murthy, Dmytro, Ramkumar Vadali and Paul Yang.
Facebook's operations and users generate more than half a petabyte of data each day, which is analysed by more than 1,000 Facebook personnel, mostly by using the Apache Hive query engine.
Typically, analysis jobs running on Hadoop are scheduled through the MapReduce framework, which breaks jobs into multiple parts so they can be executed across many computers in parallel.
Facebook ran into issues using MapReduce, however. The scheduler could not keep all the computers supplied with work. "At peak load, cluster utilisation would drop precipitously due to scheduling overhead," the blog stated.
Another issue with MapReduce is that the software typically delayed jobs before executing them, the Facebook team said. In addition, the framework offered no easy way of scheduling non-MapReduce jobs on the same cluster, and software upgrades required system downtime, which necessitated stopping jobs that are then being executed.
Facebook engineers developed the Corona scheduler so it would not have these limitations. The software would scale more easily and make better use of clusters. It would offer lower latency for smaller jobs and could be upgraded without disrupting the system.
Facebook is now in the process of moving MapReduce workloads onto clusters equipped with Corona. Initially, the social networking company deployed the software on 500 nodes. Once Corona proved effective, it was then tasked with all non-mission critical workloads, including larger workloads involving 1,000 or more servers. Now, the company is deploying Corona for all Hadoop workloads.
In tests, Corona has shown itself to be more effective than MapReduce across a number of metrics, Facebook asserted. In performance tests, Corona took around 55 seconds to fill an empty workspace, whereas MapReduce took 66 seconds - which constitutes a 17% improvement. Job are started more quickly now, as well, within 25 seconds, down from 50 seconds with MapReduce.
Corona is not the only alternative to MapReduce. Facebook also looked at Yarn, which is Apache's overhaul of MapReduce, planned for release as MapReduce 2.0. Facebook engineers were unsure Yarn could execute jobs as large as those of the social networking site, however.