TritonSort

Abstract

We present TritonSort, a highly efficient, scalable sorting system. It is designed to process large datasets, and has been evaluated against as much as 100TB of input data spread across 832 disks in 52 nodes at a rate of 0.938TB/min. When evaluated against the annual Indy GraySort sorting benchmark, TritonSort is 66% better in absolute performance and has over six times the per-node throughput of the previous record holder. When evaluated against the 100TB Indy JouleSort benchmark, TritonSort sorted 9703 records/Joule. In this article, we describe the hardware and software architecture necessary to operate TritonSort at this level of efficiency. Through careful management of system resources to ensure cross-resource balance, we are able to sort data at approximately 80% of the disks’ aggregate sequential write speed. We believe the work holds a number of lessons for balanced system design and for scale-out architectures in general. While many interesting systems are able to scale linearly with additional servers, per-server performance can lag behind per-server capacity by more than an order of magnitude. Bridging the gap between high scalability and high performance would enable either significantly less expensive systems that are able to do the same work or provide the ability to address significantly larger problem sets with the same infrastructure.

WiscSort: External Sorting for Byte-Addressable Storage

“…External sorting refers to a class of algorithms capable of sorting large amounts of data that do not fit the available memory. Among various external sorting algorithms, external merge sort is a popular approach [26,47,57]. External merge sort sorts data in two phases: run generation and merge.…”

BSP vs MapReduce

“…Use of the MapReduce framework for sorting was first discussed in the original paper [8]. Hadoop has been used in sorting benchmarking 1 [20,21,22] with very good results, when sorting 1 Terabyte and 1 Petabyte of data. Google have also independently achieved very good results sorting large data [7].…”

Sailfish

“…Our work addresses this aspect and our results demonstrate substantial gains. TritonSort [24,25] was developed for doing large-scale data sorting. Recently, in an effort that parallels our Sailfish work, the TritonSort architecture has been extended with a Map-Reduce implementation called ThemisMR [23].…”