Computing in the RAIN: a reliable array of independent nodes

Abstract

AbstractÐThe RAIN project is a research collaboration between Caltech and NASA-JPL on distributed computing and data storage systems for future spaceborne missions. The goal of the project is to identify and develop key building blocks for reliable distributed systems built with inexpensive off-the-shelf components. The RAIN platform consists of a heterogeneous cluster of computing and/or storage nodes connected via multiple interfaces to networks configured in fault-tolerant topologies. The RAIN software components run in conjunction with operating system services and standard network protocols. Through software-implemented fault tolerance, the system tolerates multiple node, link, and switch failures, with no single point of failure. The RAIN technology has been transfered to Rainfinity, a start-up company focusing on creating clustered solutions for improving the performance and availability of Internet data centers. In this paper, we describe the following contributions: 1) fault-tolerant interconnect topologies and communication protocols providing consistent error reporting of link failures, 2) fault management techniques based on group membership, and 3) data storage schemes based on computationally efficient error-control codes. We present several proof-of-concept applications: a highly-available video server, a highly-available Web server, and a distributed checkpointing system. Also, we describe a commercial product, Rainwall, built with the RAIN technology.

A group membership algorithm with a practical specification

“…In particular, an implementation of a very simple protocol for this kind of local failure detectors is reported in [12] and [36] and its state machine description is depicted in Fig. 4.…”

“…These results built a gap between group membership protocols that work in real systemsÐ-some of which have been around for many years, anticipating more theoretical resultsÐand the formal specifications that they satisfy. Examples of such real systems that run group membership protocols are Amoeba [33], [52], Isis [11], Transis [23], Totem [47], Horus [53], Relacs [5], and more recently Phoenix [40] and RAIN [12]. Researchers developing these systems are, of course, aware of the original impossibility result [28] and of its potential application to their membership protocols [17], but they also believe that their systems can work under assumptions that can easily be verified in practice.…”

An efficient XOR-scheduling algorithm for erasure codes encoding

“…RDP [10] and X Code [30]). Although there are storage systems based on both types of codes, the XOR codes outperform the others significantly [25] and form the basis of most recent storage systems [7,9,18]. This paper addresses the issue of optimizing the encoding performance of XOR codes.…”

“…Recently, both academic and industrial storage systems have addressed this issue by relying on erasure codes to tolerate component failures. Examples include projects such as OceanStore [20], GFS [12], RAIF [18], and RAIN [7], and companies like Network Appliance [22], HP [29], IBM [13], Cleversafe [9] and Allmydata [2], employing erasure codes such as RDP [10], B Code [31] and Reed-Solomon Codes [27,6].…”

A CLOUD STORAGE PLATFORM IN THE DEFENSE CONTEXT - Mobile Data Management with Unreliable Network Conditions

“…There are standard techniques for caching information on a local device (Halinger and Hohlfeld, 2010) (Nguyen and Dong, 2011) and accessing information remotely when needed, but they typically assume a reliable, high bandwidth network connection between devices (Bohossian et al, 2001) (Rhea et al, 2001). Other techniques try to deal with slow networks (Suel et al, 2004) (Shinkuma et al, 2011), but do not use the domain knowledge of applications that use this storage.…”