An Eye on the Elephant in the Wild: A Performance Evaluation of Hadoop’s Schedulers Under Failures

Abstract

Abstract. Large-scale data analysis has increasingly come to rely on MapReduce and its open-source implementation Hadoop. Recently, Hadoop has not only been used for running single batch jobs but it has also been optimized to simultaneously support the execution of multiple jobs belonging to multiple concurrent users. Several schedulers (i.e., Fifo, Fair, and Capacity schedulers) have been proposed to optimize locality executions of tasks but do not consider failures, although, evidence in the literature shows that faults do occur and can probably result in performance problems. In this paper, we have designed a set of experiments to evaluate the performance of Hadoop under failure when applying several schedulers (i.e., explore the conflict between job scheduling, exposing locality executions, and failures). Our results reveal several drawbacks of current Hadoop's mechanism in prioritizing failed tasks. By trying to launch failed tasks as soon as possible regardless of locality, it significantly increases the execution time of jobs with failed tasks, due to two reasons: 1) available resources might not be freed up as quickly as expected and 2) failed tasks might be re-executed on machines with no data on it, introducing extra cost for data transferring through network, which is normally the most scarce resource in today's data-centers. Our preliminary study with Hadoop not only helps us to understand the interplay between fault-tolerance and job scheduling, but also offers useful insights into optimizing the current schedulers to be more efficient in case of failures.

Enabling fast failure recovery in shared Hadoop clusters: Towards failure-aware scheduling

“…Thus, having long running tasks, despite the failure injection and detection times, is common in Hadoop clusters. Therefore, Hadoop is vulnerable to incur serious performance degradations under fail-stop failures [10,22,23]. Our results demonstrate that Chronos recovers to a correct scheduling behavior within a couple of seconds only and reduces the job completion times (i.e., improves the performance of MapReduce jobs).…”

Enhancing Hadoop distributed storage efficiency using multi-agent systems

“…However, striking a harmonious balance between optimizing data placement and accommodating real-time data access patterns may present its own set of challenges. In the wider landscape of cloud computing data centers, an exploration has been conducted concerning the optimization of software architecture by integrating Hadoop and MapReduce [23]. This research offers valuable insights into the interplay between these frameworks and cloud computing.…”