33069329

Reliability Issues in NoSQL Sources Lior Okman Deutsche Telekom Laboratories for Ben-Gurion School, Beer-Sheva, His home country of israel Nurit Gal-Oz, Yaron Gonen, Ehud Gudes Deutsche Telekom Laboratories by Ben-Gurion University, and Dept of Pc Science, Ben-Gurion University, Beer-Sheva, Israel Jenny Abramov Deutsche Telekom Laboratories at Ben-Gurion University and Dept details Systems Eng. Ben-Gurion University or college, Beer-Sheva, Israel Abstract—The the latest advance in cloud computer and distributed web applications has created the requirement to store massive amount data in distributed sources that provide large availability and scalability.

Recently, a growing number of businesses have followed various types of non-relational databases, commonly known as NoSQL directories, and as the applications that they serve emerge, they gain extensive marketplace interest.

These new data source systems are certainly not relational by definition and for that reason they do not support full SQL functionality. Additionally, as opposed to relational databases they will trade consistency and to safeguard performance and scalability. While increasingly hypersensitive data has been stored in NoSQL databases, reliability issues become growing concerns. This newspaper reviews a pair of the most popular NoSQL databases Cassandra and MongoDB) and outlines their particular main protection features and problems. Index Terms—NoSQL, Protection, Cassandra, MongoDB, I. ADVANTAGES The latest advance in cloud computing and given away web applications has created the necessity to store wide range of data in distributed databases that provide high availability and scalability. In recent times, a growing number of corporations have used various types of non-relational sources, commonly referred to as NoSQL sources and as the applications that they serve emerge, they received extensive marketplace interest. Several NoSQL sources take several approaches.

Their particular primary benefit is that, in contrast to relational directories, they manage unstructured data such as papers, e-mail, media and social websites efficiently. The normal features of NoSQL databases may be summarized as: high scalability and dependability, very simple info model, quite easy (primitive) problem language, not enough mechanism to get handling and managing info consistency and integrity restrictions maintenance(e. g., foreign keys), and almost zero support pertaining to security on the database level. The CAP theorem released by Joshua Brewer [1], identifies the three houses of shared-data systems specifically data onsistency, system supply and tolerance to network partitions. The theorem [2] states that only two of these kinds of three houses can be concurrently provided by the system. Traditional DBMS designers possess prioritized the consistency and availability houses. The go up of large web applications and distributed info systems, the actual partition-tolerance house inevitable, as a result imposing give up on possibly consistency or availability. The primary promoters of NOSQL sources are Web 2 . 0. 0 companies with big, growing info and system needs just like Amazon and Google. The Dynamo technology developed capital t Amazon [3] and the Bigtable distributed storage space system produced at Google [4], have encouraged many of present NoSQL applications. In this conventional paper we evaluate the security complications of two of the most popular NoSQL databases, particularly: Cassandra and MongoDB. Cassandra [5] is known as a distributed storage system intended for managing very large amounts of organized data spread out across a large number of commodity computers, while providing highly offered service with no single point of failing. Cassandra should run on top rated of an system of hundreds of nodes. Only at that scale, components fail generally and Cassandra is designed to survive these failures.

While in lots of ways Cassandra resembles a data source and stocks and shares many design and rendering strategies therewith, Cassandra does not support a complete relational data model, instead, it provides customers with a simple data style that facilitates dynamic control of data layout and format. Cassandra was created to support the Inbox search feature of Facebook [6]. As such it can support over 75 million users which use the system continuously. MongoDB [7] is a document data source developed by 10gen. It handles collections of JSON-like documents. Many applications can thus model info in a more normal way, as data can e nested in complicated hierarchies but still be query-able and indexable. Documents happen to be stored in choices, and choices are in return stored in a database. A series is similar to a table in relational DBMS, but a series lacks any schema. MongoDB also supplies high supply and scalability by using Shardings and Imitation sets (see below). The increasing demand for NoSQL sources such as Cassandra and MongoDB and the huge amounts of userrelated sensitive data stored in these types of databases boost the concern for the privacy and personal privacy of the data and the secureness provided by these systems.

With this paper all of us review the main security features and challenges of these two database systems. We begin with a brief introduction to Cassandra and MongoDB functionality in section II. We all then discuss protection features of Cassandra and MongoDB in parts III and IV correspondingly. We consider in section V. Since much of the dialogue is based on open-source Internet documents, it obviously reflects the case at the time this paper is written 2011 International Joint Conference of IEEE TrustCom-11/IEEE ICESS-11/FCST-11 978-0-7695-4600-1/11 $26. 00 © 2011 IEEE DOI 10. 1109/TrustCom. 2011. seventy 541

Tweet