You are on page 1of 6

NoSql Database

Abstract-The relational database has been exponential growth of data ,since many

the dominant model for database organizations collect vast amounts of

management since it was developed. customer, scientific, sales, and other data for

However, organizations that collect large future analysis. Similar issues are involved

amounts of unstructured data are in modeling the social network like Twitter,

increasingly turning to nonrelational Facebook, etc. Implementing such problems

databases, now frequently called NoSQL in relational databases involves large

databases. NOSQL databases are distributed, number of joins which is expensive to be

non-relational, open source and are calculated. With the intense increase in

horizontally scalable. In this research paper, usage of internet leading to need for storing

we are surveying about NoSQL, its large amounts of interconnected data, there

background, features and various NoSQL was a clear desire for a data store tailored to

data stores. such needs.

I.INTRODUCTION Thus a piece of software is a NoSQL


database if it doesn’t require a stringent
Relational model has been dominating the
schema for every record created, is
computer industry since the 1980s mainly
distributable on commodity hardware &
for storing and retrieving data. With the
doesn’t use relational database mathematical
widespread use of relational database
theory. The primary advantage of NOSQL
systems the flaws of such systems become
Database is that, unlike relational databases
more apparent. Way back people used
they handle unstructured data such as
database just for storing tabular data like
documents, e-mail, multimedia and social
purchase reports and finance records &
media efficiently. NoSQL databases aren’t a
relational databases were perfect for this.
mere layer on top of existing technologies.
But in today’s environment, these relational
Each type of NoSQL database is designed to
representations are not efficient in
manage different types of data. The
performing many operations. For example
popularity of NoSQL databases lies largely
the World Wide Web exhibits far more
in the speed they provide for developers.
complicated networks of relationships than
NoSQL solutions are quicker to build,
were expected when SQL was designed
update, and deploy than their relational
.Also Relation database lacked handling
forerunners. sharing has dramatically increased. Much of
II.BACKGROUND this data is unstructured text.

The first documented use of the term There were problems with conventional

NoSQL was by Carlo Strozzi in 1998. Carlo databases. Not every problem fits well into

used the term NoSQL because his database rows and columns. There are also scenarios

was accessed via shell scripts, rather than in which the relationships themselves can

through use of the standard Structured hold their own metadata. An RDBMS

Query Language (SQL).Thus the name, doesn’t handle those situations at all. Every

NOSQL. Its origin can also be related in scenarios has a type of NoSQL database that

2006 when Google released its paper that overcome the limitations of RDBMS.

described its Bigtable distributed structured Relational databases can suffer from a

database. Amazon also released a paper of sparse data problem .We may discover facts

its own in 2007 describing its Dynamo data and relationships over time. Using an

storage application. The Dynamo paper RDBMS for this would require an

basically described the first globally ever‐increasing storm of many‐to-many

distributed key‐value store used at Amazon. relationships and linking tables.

These two papers inspired many different IV.NOSQL FEATURES


organizations to create their NoSQL Schema agnostic: A database schema is the
databases. Thus, many open-source NoSQL description of all possible data and data
databases had emerged by 2009. Riak, structures in a relational database. With a
MongoDB, HBase,Accumulo, Hypertable, NoSQL database, a schema isn’t required,
Redis, Cassandra, and Neo4j were all giving you the freedom to store information
created between 2007 and 2009. Today the without doing up‐front schema design. The
NoSQL movement includes hundreds of great benefit to a schema agnostic database
NoSQL database products. Each problem is that development time is shortened. In a
resulted in its own solution and its own traditional RDBMS, many times we go
NoSQL database. through a process of schema redesign. We
III.WHY NOSQL may have to modify our queries accordingly.
Due to recent trends in IT & the advent of NoSQL databases provide this flexibility for
the World Wide Web and the explosion of changing both the schema and the query.
Internet connected devices, information
Nonrelational: Relations in a database Highly distributable: Distributed databases
establish connections between tables of data. can store and process a set of information on
With a NoSQL database, the information is more than one device. With a NoSQL
stored as an aggregate i.e a single record database, a cluster of servers can be used to
with everything about the transaction. hold a single large database.
NoSQL databases don’t have this concept of Other features Although some features are
relationships between their records. They fairly common to NoSQL databases
instead denormalize data. This approach has however certain NOSQL databases may lack
the advantage of not requiring complex one or more of the following features and
query time joins across multiple data still qualify as a modern NoSQL database.
structures (tables). Rather than inserting ten Open‐source: There is not a host of
rows in a variety of tables in a relational open‐source NoSQL products to suit every
database, we can instead store a single need.
structure for all, as a JSON or XML BASE versus ACID: Prior to 2014, the
document. In relational database theory, we majority of NoSQL definitions didn’t
normalize data. The advantages to storing include ACID transaction support. Not many
data multiple times are Easy storage and NoSQL databases have ACID transactions.
retrieval & Query speed. Exceptions to that norm are FoundationDB,
Commodity hardware: With NoSQL Neo4j, and MarkLogic Server, which do
database, cheap off‐the‐shelf servers can be provide fully serializable ACID transactions.
used. Adding more of these cheap servers V.MANAGING DIFFERENT
allows NoSQL databases to scale to handle DATA TYPES
more data. In many NoSQL databases, a key
NoSQL databases come in four core types.
design decision is to use multiple computers
one for each type of data the database is
to store data for a single database, rather
expected to manage:
than have the whole database on a single
Columnar: It is an extension to traditional
server. The main advantage of this approach
table structure. Supports variable sets of
is in the case of very large datasets. Another
columns (column families) and is optimized
key advantage is that adding high
for column‐wide operations (such as count,
availability is easier;
sum, and mean average).
We can define column families before Key-value Store databases are more
loading data into the database, meaning that appropriate for the management of stocks,
the structure of data must be known in products and real time data analysis,
advance. providing high data retrieving speed while
Each one of the column families consists of the greatest amount of data is mapped into
several fields. One of these column families memory.
may have multiple “rows”. A key benefit of Basho Riak, Redis, Voldemort, Aerospike,
a column store over an RDBMS is that Oracle NoSQL are few key-value stores.
column stores don’t require fields to always Triple or graph store: It is now becoming
be present and don’t require a blank padding prominent; however the concept of triples
null value like an RDBMS does. This has been around since 1998. Every fact (or
feature prevents the sparse data problem. assertion) is described as a triple of subject,
The great thing about column stores is that predicate, and object. We can quickly build
you can retrieve all related information this simple data structure into a web of facts,
using a single record ID, rather than using which is called a directed graph in computer
the complex Structured Query Language science. Graph math is complex and
(SQL) join as in an RDBMS. specialized and may not be required in all
Column Family databases should be used situations where storing triples are required.
when the number of write operations If we need to store facts, dynamically
exceeds reads. changing relationships, or provenance
DataStax, Apache Cassandra, HBase, information, then we consider a triple store.
Apache Accumulo, Hypertable are few If we need to know statistics about the
columnar databases. graph, then we should consider a graph
Key‐value: It has very simple structure store.
having sets of named keys and their Graph databases are more appropriate for
value(s), typically an uninterpreted chunk of working with connected data, for example,
data. Sometimes that simple value may in to analyze social connections among a set of
fact be a JSON or binary document. This individuals, road maps and transport
structure is also known as “hash table” systems.
where data retrieval is usually performed by Neo4j, Ontotext’s GraphDB, MarkLogic,
using key to access value.
OrientDB, AllegroGraph are few graph ACID Consistency: ACID means that once
databases. data is written, we have full consistency in
Document: XML, JSON, text, or binary reads.
blob. Any treelike structure can be Eventual Consistency (BASE): BASE
represented as an XML or JSON document. means that once data is written, it will
Document databases are sometimes called eventually appear for reading.
aggregate databases because they tend to Some NoSQL databases have
hold documents that combine information in ACID‐compliance on their roadmap, even
a single logical unit, an aggregate. Loosely, though they are proponents of BASE. Many
a document is any unstructured or NoSQL databases allow tuning between
tree‐structured piece of information. It could levels of consistency and availability, which
be a recipe, financial services trade, relates to the CAP theorem.
PowerPoint file, PDF, plain text, or JSON or The CAP theorem is a computer science
XML document. conjecture, shows the list of options as to
Document Store databases are a good choice how the balance between consistency,
when working with large amounts of availability, and partitioning can be
documents that can be stored into structured maintained in a BASE database system.
files such as text documents, emails or XML Eric Brewer came up with the CAP
and CMS and CRM systems. conjecture in 2000. The CAP theorem in
MongoDB, MarkLogic, CouchDB, computer science is fundamental to how
FoundationDB, IBM Cloudant, Couchbase many NoSQL databases manage data. CAP
are few document databases. stands for Consistency, Availability, and
Hybrid NoSQL database: These databases Partitioning, which are aspects of data
provide a mix of the core features of management in databases.
multiple NoSQL database types such as Consistency: Shows whether the database is
key‐value, document, and triple stores all in fully (ACID) consistent, or eventually
the same product. consistent, or without any consistency
VI .CONSISTENCY METHODS guarantees?

In the NoSQL world, consistency generally Availability: During a partition, is all data

falls into one of two camps: still available?


Partitioning: If some parts of the same Conference on Cloud and Service Computing , vol.,
no., pp.336,341, Dec. 2011
database cluster aren’t communicating with
doi: 10.1109/CSC.2011.6138544..
each other, can the database still function [2] Zvarevashe,K.; Gotora,T .,” A Random Walk
separately and correct itself when through the Dark Side of NoSQL Databases in Big
Data Analytics” , International Journal of Science
communication is restored? and Research , vol.3 no.6, pp. 506-509 ,June 2014.
The CAP theorem states that we cannot have [3] Okman, L.; Gal-Oz, N.; Gonen, Y.; Gudes,E.;
Abramov, J., “Security Issues in NoSQL Databases”
all features of all three at the same time. It is
Proc. of IEEE 10th International Conference on Trust,
claimed that we can have only two of the Security and Privacy in Computing and
three. The reality is that each is a sliding Communications , vol., no., pp.541-547, Nov 2011
doi : 10.1109/TrustCom.2011.70.
scale. We may be able to trade off a little of
[4] Nayak,A; Poriya, A,; Poojary,D., “Type of
one for more of another. NOSQL Databases and its Comparison with
Relational Databases” , International Journal of
Applied Information Systems issn : 2249-0868 vol.5
VII.CONCLUSION no.4, pp. 16-19 March 2013.
[5]Cattell,; Rick. , “Scalable SQL and NoSQL data
It’s tempting to think that once NoSQL
stores” , SIGMOD Record ACM , vol.39, no.4 ,pp
databases evolve to handle more data and .12-27, December 2010 .
query types, the RDBMS will no longer be [6]Jing ,H.; Haihong, E.; Guan, L.; Jian ,D., “Survey
on NoSQL database" , Proc. IEEE 6th International
needed. Nothing could be further from the
Conference on Pervasive Computing and
truth because NoSQL databases can’t Applications” , vol., no., pp.363-366, Oct 2011 doi:
provide all the functionality of a relational 10.1109/ ICPCA.2011.6106531.
[7] Leavitt, N.; , “Will NoSQL Databases Live Up to
database. We can think of Hybrid database Their Promise? “, IEEE Transactions on Computer,
model, that is, they have the familiar SQL in vol.43, no.2, pp.12-14, Feb. 2010 doi:
10.1109/MC.2010.58.
the frontend and NOSQL in the backend.
[8] Tauro, C,; Aravindh, S,;
These databases might not be as fast as a Shreeharsha,A,B.,"Comparative study of the new
pure NOSQL database but they still provide generation, agile, scalable, high performance NOSQL
databases", International Journal of Computer
features of both relational as well as
Applications issn no 0975 – 888 vol 48, no.20, pp 1-4
NOSQL databases to the user. Thus a lot of June 2012.

disadvantages of both relational as well as


NOSQL databases may be covered up.

REFERENCES
[1] Hecht, R.; Jablonski, S., “NoSQL evaluation: A
use case oriented survey”, Proc. IEEE International

You might also like