Evaluation of active storage strategies for the lustre parallel file system

Abstract

Active Storage provides an opportunity for reducing the amount of data movement between storage and compute nodes of a parallel filesystem such as Lustre, and PVFS. It allows certain types of data processing operations to be performed directly on the storage nodes of modern parallel filesystems, near the data they manage. This is possible by exploiting the underutilized processor and memory resources of storage nodes that are implemented using general purpose servers and operating systems. In this paper, we present a novel user-space implementation of Active Storage for Lustre, and compare it to the traditional kernel-based implementation. Based on microbenchmark and application level evaluation, we show that both approaches can reduce the network traffic, and take advantage of the extra computing capacity offered by the storage nodes at the same time. However, our user-space approach has proved to be faster, more flexible, portable, and readily deployable than the kernel-space version.

Oasa: An Active Storage Architecture for Object-based Storage System

“…3, the input and output of a function can be both network client and disk. Similar to the previous work 13,14 , data flows in pattern 1 and pattern 3 are as same as that in the normal OSD WRITE and READ command. On this occasion, intelligent operation such as data filter (e.g.…”

“…Since object-based storage technology may significantly influence the networked storage industry, a lot of researchers have gradually made efforts to integrate active storage into the object-based storage system. Piernas et al gave an active storage strategy implemented in userspace for Lustre parallel File System 14,10 , which proves to be faster, more flexible, portable, and readily deployable than the kernel-space version. Huston et al presented an active storage architecture called diamond 22 , used for interactive search of nonindexed data from the massive storage system.…”

“…John et al 13 proposed an active storage framework for object-based storage device. Piernas et al 14 gives an active storage strategies implemented in user-space for Lustre parallel File System 10 . Qin et al 15 present a hybrid approach combining a requestdriven model and a policy-driven model to execute code on OSDs.…”

IOStack: Software-Defined Object Storage

“…IOStack borrows ideas from active storage literature [13,12] as a means of building filters to enforce policies.…”

In-vivo Storage System Development

“…Previous work on active storage systems have shown the benefits of co-locating processing with data, such as reducing data transfer and exploiting I/O and CPU parallelism [1][2][3][4]. And more recently, the Rhea [5] system showed that filtering kernels could be extracted from Hadoop jobs using static analysis and applied transparently in cloud storage services such as Microsoft Azure [6].…”