ISSN No.

0976 -S69 7 ]
DOh http1/ /da.do l.w9/1 0,Z4M H/llcw c•.v81 a•
.sn
Volu me 8, No. I, Sep tem ber- Octo ber 2017
€\ lnte mat lona l Joum al of Adv ance d Research In
Com pute r Selene•
RESEARCH PAPER

\~ Ava ilab le Online at www.lJarcs.lnfo

NTR OL MO DE L WI TH
A SEC UR E AND ROBUST CO NG EST ION CO
BUTED DATABASE SYS TEM
FRAG1\1ENTATION FOR CLU STE RED DISTRI
Surya Pratap Singh
Arvind Kumar Maurya Department of Computer Science
Department of Computer Science DOU Gorakhpur University
DOU Gorakhpur Univmity Gorakhpur, India
Gorakhpur, India

Upendra Nath Tripathi

Manish Mishra Department of Computer Science
Dcpanment of Electronics DOU Gorakhpur University
DOU Gorakhpur University Gorakhpur, India
Gorakhpur, India
ue S)'llem (008 S). Distn buted ~ubu e ii
t a aecure and robust way 10 adopt Distributed Datab
A6stN tt: The purpoiC of this paper is to presen data b~ ~ c:he n~~
buain eu environment bu III iacreu inaJy need for distrib uted
bcc:omina very popular now a day. Today"• le, reliab le and accessable anformabOn u.
oloaia ia to fulfil the desire for coaaia lcnt, acalab
applications. We know that the aim of cumn t techn datab ue aite'a cluste rina ~la m are NP-
ue System, the data &apnentatioa and distributed
steadily powi na information. In Distributed Datab of data means , while uppadina from Rclabona.l ~aub ue
pointa c:omidered dud may cause lou
Hard in nature and difficult to aolve. One of tbe a lo acccsa the applic:ation1 on the dala-1 1ta .
•
UICI hpne ntatio n. 1be datab uc queri
System, data distribution over clift'ermt data-aitca a are neede d lo be alloca ted to tbc distrib uted datab ue Illa.
frqmc nta lbat acceucd by queri
should be performed ctr~v cly. Therefore, the d one may ctusc expcn siven cu oftcch nolos :Y
be u accurc u data be loule u durin& query execution. And, the secon
Thus, fi"agm cn11tion should
Diltr:ibuted Datab ase Systcma in order to achie ve not only NCUN
na the aitca (clustcrina) of
chana es. Thus, we present a method for poupi
stion control.
fi-agmentation but also robust and coat effective conae
Cluste ring nctWOrk ailcS.
stion control, Rcpo1itory clala-sita, Frqmcntation,
Kr, ~s: Distributed Databue System, Robust conge
added i.e.
datab uc aystems. However, a new aspec t is
I . INTR ODU CTIO N to do distri butio a
distribution transparency. In order
layered refere nce arc:hi tcc:tu re
were transparency, wc defm c
It is well known that durina the 1970 . computcn This archi tectur e provi des a very pncr al
powe rful, intear aled datab ase (fi,ur e-1).
extensively used for buildina
olol)' of datab ase system s has built ill conceptual framework.
systems. The tcchn ate the
found ations and has been exper ience d in a larae The three most important objectives which motiv
theoretical features of tbil arcbit ccture are:
number of applications. At tbc same time,
computer
of data
networks have been exten sively devel oped allow in1 the • Location Transparency, the separation
of different comp utcn and excha nae of data and fiapnentation and allocation
COMection dancy,
other resources betw ffn them. • Replication Transparency, the control of redun
uter
So for, the availability of databases and comp and
ce &om
networka has Jivcn rise new field: Distri buted Datab ase • Frapncntatioa Transparency, the independen
buted Datab ase local data sitca.
System. In order to understand Distri that aJao
Systems, we mua& know more than the princi ples tradit ional Apar t from this issue. we observe another one
atc ~ect a ~ adaptation o( tbia tec:bnolo11 cbu p1 i.e. coat
databua and computer networks. We must also iatca,the a coat- effec tive
tbil knowled1e with the study of the particular upcc: t of mcff'cctjvencu. Here, thus, wc consi der
a) problem of data 1ite1. To reduc e tbe
new tecbaolo,y. The moat important tccbnoJosic tecbniquo i.e. cluatc rin1
rn distri buted
of disttibulecl databases derive s from the issues 'co- upcn sive cban p from traditional one to mode
operation betwe en auton omou a sites' occur red. environment cluatcrin1 provides dominated featurea.
buted
Distributed Database Systems arc not 1imply distri
ase system s. 8ccaU le of'
implcmcn1ation of centralized datab
of syste ms which prese nt dift'cn mt
they allow the daip
a. There ia
rcaaur. fiom tnditi onal. centra lized aystcm
buted Datab ase
varioua upec t of' data security in Distri
layer.
Systema auch • ,ecwity II coaccptual and application
ity can be iq,lem ented by Illini
The application layer aecur
ol polic ies. Here. wher cu, we only conce mod
acceaa coalr
with conceptual layer aecurity iaua .
Very fine. ia Diacn"bulecl Dllta bue System. we ob1a
w
die impo rtance U in lrlditi onaJ
dala independence hu 111111
'·
q -•
1 , , \\7 •.
I ',., I ... ,, •• , ,, , , . , " " I ' I " I . ' ... , , ..........., I •' " ' " 11.,1 ..r \th ,1111, .1 knc•" h ... f"fltnpulrr , , irnc,, II f lll, ,. ,,, ' h I 2 • -

But. there i1 another way a relation can be fraamc~tcdd

i.e. Mi•cd or Hybrid fra1mcnc.1ion which can be obtaine
by 11.•ina the above both fnamcnta11on1. .
Horizontal Fra~"'""tatiori i1 the 1ub~t of the rct.-u~n•
-., that aroups all the tuples in accordance with • tc~c:chon
condition on values of one or more fields. Each horizontal
fragment must have all columns of the original bL'IC rclallOn
and should also confirm to the rule of re-c:onstNCtivcncss.
Whereas, the Jl,rtical Frog,,,,ri,ation of a global relation
i1 the subdivision of ill fields or columns that are aroupcd
into fragmcnll. In order to maintain re-c:onscructiveness,
each fragment should contain the primary key field(1).
Vertical fragmentation can be used to enforce privacy of
-
0
GI
GI
·--- ---·
'I Inc-al •t
data.
Corrttt11ft6 R11ln ofF,.,,,.,,.,.d on[ I ):•
~ : Os-•h"C:
an !,"''""· C""'Pl,tntUI.
~

"'"'0
"···r····
I I I

Recon.rtr11ctlon
Disjoint,wu
~
~ A,N11t•1n ofFNp,•ado 11: •
• Since data is stored close IO the local site of usage,
efficiency and performance of the database system is
Figure I .A reference architecture for DDBS
incnased.
• Local query optimization techniques arc sufficient for
most queries since data i1 locally available. Thus, easy to
2. FRAGMENTATION S\'NOPSIS
manage balanced storage capacity and cost.
• Since irrclcvanl data is nol available at the local sites.
Fragmentation is a design processes that allows dividing
aecurity and privacy of the database 1y1tem can be
• single relation or c:lus of a database into two or more
maintained that increase the reliability.
partitions. Fragmentation must be in such way that the
DIHt#N11t•1n of F,.,,,.,.,.do 11: •
condition of the panitions provides the oriainal databuc
without any loss of data and each fragment can be stored at • In the cue of recursive fragmentations, the job or
reconstruction will need expensive practicet.
any data-site over a computer network of a Distributed
Databue system. • Lack of back-up copies of data in differcnl local lites
FN1•fflMll 011 .... ,. i,,,pr,,w:- may render the dallbue ineffective in the case of failure.
- Reliability • When data &om different &agmenll are required, the
- Security ICCCII speeds may be very high.
- Performance
- Balanced 1tonge capacity and colll J. CONGESTION CONTROL SYNOPSIS
- Communication costs
As we know, 11 the top level of 'reference architecture The demand for more and more information both by
for Distributed Database System' i1 the global schema. The industry and aovcmmeat leads to databases that will exceed
Jlobal schema consilll of the definition of a set of alobal the physical limitatiooa or centralized 1y1tcms. Thus, with
relations. And. each alobal relation can be 1plit into sevenJ the rise of user bued oa Distributed Databue System.
non-overlappina portions which are called hgmentations. traffic congCllion ii one of the unavoidable situations. Even
The mappina between alobal relalions and fnamcnta i1 though several mcan:hcn addrcu the congestion detection
defined in the f'rapnentarion schema. This mappina ii one to technique, ill avoidance ~ mitigation in their racarcb are
many. Fraamcnts are loaical portions or alobal relations. hard to be explored for any eff'ectivc solution for this
Mc:ana, fraptcnll are physically located a& one or ICVCrll problem. Conpon can be outlined u a condition that
sites of the network. And, lhe •llocotiOII «l,nu, definet at happens once the network mourca are overburdened. So
which 1itc(1) • traamcnc is located. At the data procasina raullin~ in such i11ucs is measured by 1011 or data or delay:
level. it i1 required to map lbe physical imaaes to the local Con,auon control 11 • fundamental mechanism u a
DBMS.. This mappin1 is called • local mappina schema troublcsbooter of such problems. Conaation manapment
and depends on the type of local DBMS. So, in a could be reflected u alaorilhm to shire network reaourca
hcteropneoul l)'llem. we haw differcnc types or local amon1 competitiw traffic IOW'Cca. However TCP
mappin1 at different 1ita. con,estion control alprithms can be interprded •
distributed primal4111 alpithms over the lnlcmd to
muimize aarepte utility. Hence, there is a need or lhldia
Tbe dccompolilioa or alobal relatioas into hgmeata can that cxcavatca the loopholes in various approacha 111d
be performed by applyina two different types of lllilll to identify • desip of new conpllion COlllrOI
hp,ca -. (11(2): ledmJque, Coa,atioa IIIIIIIICfflClll mechanisms la loday"I
• Horimalal Frapeataricwl. ucl lntcmct enormously typical to implement Becw. lbe
Internet continuea to expand in size. diversity, 111d reecll.
• Vcnical frllft't aPlliOa
takint put in aa evcr-increaaina role within lbe • . - -
of diff'creal aetworu (lrlalpOrtati on,, finance. etc.). II ii ID
 I I,,,

t,c noted that J1stnhuteJ network 5ystem mu~ not be the conse~tton window for e,,ery TCP ,uh-now 1hcrefore on
thought of u •~'<l\:iale uniting of elc,tronic network Sikh•• m1t1g1t1n11hc variety of cnd-1o-cnd r••I 11 1dJre\\C,
h dcay.
the internet. Hence. \\f arc ,o,ng to J,~u...s the r1rt1cul..r both -.'irtd and wireleo nttwork,.
Jistnhutcd network 11>·r-1cm that may he 1JJ1t1onally
cl.t.,~ificJ into a ,cry l.art:c nctv.ork of internet. v.irclcu 5. cu:ST[RISG NF.T\\'ORK ~nr.s
~ensor nct-wurl. and muh1lc ad-hue nctv.·ork. . • "a
First, we must undcntand •hat du,tcrin1 " ·
4. l.lT[RATl"R[ RE\'l[W groupina of rclalcd items stored to1cthc:r fur dfi"cncy of
access and re,ource utilization".
Many stuJ1es ha,c hccn publu,hed on attempts of The raramctcn con,idcred for the propoted chL1tcnn1
1mrro,1n~ the rcrformance of ODDS. These researches technique are described as follows:
ha, e mostly in, estigated fragmentation.. allocation and • Logical Cliut,r C, Lo1ic1I place 1h11 uwd IO sroup
~omcumcs clU5tering problems. In this section, v.·e present network sites together based on some physical property
the main contnbuuons related to Clw.tercd approach lo like di5tancc between them exist.
Congcst1on Control v.1th Fragmentation. • Distrihut,d Networlt SiteJ Set of fully connected nctWork
The authors of (3)14) rrcscnt a new formulation for the sites S., s, , ... s. of distributed dau~ase sy~tei:n, Each
rroblem of fragmentation and allocating those fragments site is the place &om where lhc transaet1ons arc tnuercd.
v.1th minimum cost for both structured and unstructured and transaction results arc held.
data, by ~uping sites which are nearer lo each other into • Distance Range DR The maximum distanee ,·aluc (tn
one clu.-.tcr, hence they ha,·e low cost. Also, a dynamic Km) that is allowed between the DDS nctw~k sites for
clustering method is adopted for both structured and pupina into the umc cluster can be decided by the
un~ttuctuttd database to reduce the movement of data network administrator. Shortest path method is used lo
~-ccn sites. ulculate the distance between two sites.
The complCJtity of a distributed database algorithm • Distance D(S,,S.,) The shortest path distance between two
depends on the allocation method used. Some enhancements sites S, and Si in the DDS. . ·.
have been done in reallocation algorithms. The authors of • Cluster Site Matrix - CSM Calculated malnll by which
(6) proposed an algorithm that reallocates fragments based the clusters arc created and their network sites arc
on the distance between sites to minimize the number of assipcd.
communications and network overhead between sites and • Clustering Decision Yalue - CDY The binary value that
also calculate the cost for each fragment individually. The dctcnnincs whether a pair of sites S, and SJ can be
reallocauon depends on finding the maximum update cost p-ouped together in the same cluster. CDV is calculated
\'Blue for each fragment. This technique takes into account using following fonnula:
the nctWork topology and set of queries &cquency values 1 : ,, D(Si,Sj) S DR AL ._ J
employed over the netWOrk. CDV(Si,Sj) = { 0: ,, D(Si,Sj) > DR VI• J
The authors of [7] present a bio,copphy-bucd II is obvious that CDV for the same site is equal to zero.
optimization technique for no-replicated ~ allocation of If the CDV(S,. SJ) is equal to I, then sites S,. S, arc grouped
data fra,mcnts durin1 database desian tha1 minimize total into the same cluster, otherwise they arc assigned to
data transmission cost durin1 the execution of a set or different clusters. Suppose the seven sites of distnl,Uled
queries. database are placed at some distance (in Km) &om oac
We, now, discuss the mosl iq,ortan1 work dispemcd lo another accordin1 lo the site distance matrix shown in the
addrcu congestion controls in nccworkin&- Scferoglu et al. following table.
(9) hau· conferred (mdinp on TCP-induced packet loua of Table-I·- Site distance matrill
TFRC (TCP Friendly Rate Control) flows and their relation SIN SI sz SJ s, SJ s, S1
•ith the delay umpla and their derivatives as SI 0 70 290 ISO 340 S20 180
collccted'coq,u&cd at TFRC scndcn and receivers. Mao et sz 70 0 640 260 72S 400 S40
al. ( I OJ have developed a hybrid lraffica Aciive Queue SJ 290 640 0 80 60 270 830
Management (AQM) tOUICr with classifier and scheduler
dial make sure die link capacilia of each &raffle. The aulhor
s, ISO 260 80 0 60 SIO 420
SJ 340 72S 60 60 0 3SO 610
has conferred IOIIIC easily verified adequalC stability
conditionl for die AQM policy lo atabiliu the TCP and
s, S20 400 270 SIO 3SO 0 so
UDP queues in routcn- Shianl and Schaar [ 11) have S1 180 S40 830 420 610 so 0
propotcd a cantn-ewarc conpation 1111111,cmenl for
multimedia ,ysccm .aramin1 over TCPnP networks For ICltinl up an efficient clustcrin& mccbod,, ii is
achicvinl hiper than JdB i1■povcmca& ia tcrma of PSNR auumcd that cacb lite ii usiped to only one cluster. ne
over die lrlditianal TCP con,ation COlllrOI approaches. cl111tcrin1 alaoritbm is described u follows:
,,,,,,,,.,,,, ,Wtm,
widt the .,__ a,hanccmenll clilCOVCNd for real-rime
1tra111ina applicaf-GIII requiriDa ripoua playback delays.
,,,,,.,,:
bhmla cl al. (12) ,_. inaoduced a proxy lnnlport layer Site Diltancc Malrix ahowia& distance bet111eaa lilel
prC110CG1 .,...._ Cqrdioe Control Pnleocol (DCCP) DtS..S,)
Dimnceaan,e(DR):
dlll't app•illl for dlllC applicalionl clue IO ica cxcluaive
..,_,,.._ Z11o1111 al. (13) haw prwnted • conpslion Number of IIIWOlt lita or di11rtbuled dllllJIN . , _
(NS)
.,_,.■ ea..:. ......~~.. ~~-~Ml~pach
Tt•P• ~ PtulDCDI) ......,., - ..,--Y .uJIIIII
 . . o 1017 H7-.l4l
In C11111pulcr s,·,rnct.111111. Sc-pl c1 • •
, , • " " "11 ,,, 11 'l.1111, A , , 11/, I"" r11~111111:il .lourn:11 nr .\1h 311n·cl kc,1"11rch

Procf'ssm g:
this different vulnerabilities may exist due to heterogeneities
in computational units. .
1•oe1ermining the sires that match the distance range in
order to group them in one cluster-/ We start to address security heterogeneity 1ss~cs by
Slep I: Set i• I dividing a big storage data-site into several data-s11es or
Step 2: Do steps (J-12) until i>NS different type groups. Each data-site type rcpmen ts a leve
st
Step J : Set j• 1 of security vulnerability. In a data-site type group, orage
sites with the same vulnerability share the same weakn~s s
Set 0 to cluster site matrix CSM s' informal lon
that allows attackers to reduce the data-site
Step 4: Dosteps (S-I0)u ntilj>N S
Step S: lfi t j AND D(S" SJ)<• DR, assurance. Although it may be difficult to classify all d~ta-
go to step (6) Else, go to step (7) sites in a system into a large num~ of grou~s, a practi~I
Step 6: Set I to the CSM (Si, Sj), way of identifying data-sites types IS to organize these wilb
go to step 8 similar vulnerabilities into one group. In real-world
Step 7: Set Oto the CSM(S,, S1) distributed systems, the fragmentation technique is usually
Sttp 8: End IF combined with replication to achieve better pcrform a~e at
Step 9: Add I to j the cost of increased security risk to data stored in ~
Step I0: loop systems. A practical distributed system normally contains
Step 11 : Add I to i multiple heterogeneous data-sites providing services. ~th
Step 12: Loop various wlnenbilities. Unfortunately, the elUslmg
Output: Cluster Site Mattix (CSM) having generated fragmentation algorithms do not take the heterogeneity
clusten and their respective network sites issues into accounL Our fragment allocation solution we
End proposed in this paper is a bit different from the existing
Suppose Distance Range (DR) value is 100 Km. After fralJDCnt allocation schemes discussed so far. As in lhe
using the abo\·e clustering algorithm and site distance figure-2, our solution captures heterogeneous features
mattix. a Cluster Site Matrix (CSM) is produced u shown in regarding vulnerabilities of the nodes in order to improve
the following table. the security and congestion control of the data stored in a
Table-2:- Cluster Site matrix (CSMl distributed system.
Sit, SI Sl SJ s,
SJ S1 s, Moreover these aspects, a technique of fragment
allocation and replication at clusters are evaluated accordin g
Cl I I 0 0 0 0 0
0 0 1 I I 0 0 to the performance generated by reducing the size of
C1
0 0 0 0 0 I I fragments that allocated finally at the clusters. The closest
CJ
methods in the literature to the proposed technique of
fragment allocation and replication are those proposed
6. RESEARCH METHODOLOGY
previously by some researchers. The main differences
between these two methods arc described as follows.
We observed several works related to data fragmentation
and their security in Distributed Databue System. In tbia
paper, we find the adaptation of technology chanaea from
relational database to distributed darabuc become too
expensive to use it in cost effective manner and issue of data
security in Disttibuted Database System motivate us to
apply some new to do data fragmentation. Thus, we
proposed the followina technique by which we able to apply
secure and cost effective way for Distnbuccd Databue
System.
su,a ,,,,.,,.,tfPmdn ., a,.,, ,., • ,,,.,,,,,,,.
tioa
As 1CCD above tbal hpncnta be used a base
CID U
for data located oa different aita. Now dmactcristica of
these data can be used for aecurity purpoee. A data CID if
interrupted oa one lite lbcn it needs olbcr related data be
kept oa anotbcr lite bcc8UIC actual meanin1 of it ii 111eleu
unlca lhia partiaJ clara ii aot dnaten. lflbis poaibility ii to
shorten and it ia c:omiclcred dlll only two 1ita are havin1
precioua dala lbcn allo half of lhe data ia miaiq IO dlil ia a
secure method.
----. 1-
o---.e
.,._..
c........
.....
In a larJC scale diltribufed system. diff'Cl'CIII storage sites
have a wriety of wa11 to protect data. The umc security
policy may be iq,lcmmred in varicu mcc:banisms Data Duster Stana•
cacryptioa ICbemcl may wry. eVCD witb the 111DC Sublpte m
cacryptioa ldlemc, by ....... may wry acrou the
Fipre 2.Arcbitecture of Distributed Database S11tem witb
diltnbulcd IJIICIIL Tbe lbow mentioned facton can NCUrt hpnent allocation al cl1111cn due to CODp1tioa.
coaaibulc 10 dift'cral Ylllnmbilitiel UIIODI l&Orap lites.
Ahboup NCUri'1 mocblniaw deployed in Dlhiplc data-
Ifwe co...,_.
witb tbe diff'erclll allocation tec:bniq w
introduced by IOIIII IIIIU'Cben u meatioaed ..
lites caa Ill illlph■•s1 ii • bomopDIO'II way, besides literatun review, OIi' alloc:llioa and "Plic:atioa _..., .

••,J • I , I \Ill ' \ II , If ll.-..rhJ

 117
\ 1 •• •" ' h 111 11 , , ' l .11tr , • , 111 /. I 1111111~111111JI .101,r 11:al of .\1h JIit, 11 kn, arch In ( 1,mpulu ~lrnn. ll flll, ,, pl f lcl !Ii I 7- ··'·'2
considers cost effective between cluster 1ite1 for network International Journal of Database Theory and
~ommunications such as the update and retrieval costs (i.e., Apphcation.pp.45-60. . d
II mostly rcr,rci.cnts the cost of writing operation that takes (7) Singh. A., KahJon, K.S. and Virk. R.S. "Nonrcp1ica~e
r,lace . during the . execution time). Moreover, in clustering Static Data Allocation in Distributed Databa...CS U~mg
h:chmquc, Jat.&-i,;1tcs 11.re grouped according to a clustering Biogcography-Based Optimizallon"
range and not only to a specific communication cost. Our Volume 2014 (2014), Chinese Journal of Engmeenng.
clusters can communicate with each other instead of pp.1-9
r,rcfcning to ha\'e all fragments in their sites. This (8) Ahmad, I., Karlapalern, K., Kwok, Y.K., So, S.~
communication technique is cheaper than allocating ..Evolutionary algorithms for allocating data in
fragments in all sites. In addition to the advancement in term distributed database systems" Volume
of communication cost, the indepcndcncy of our clusters 11, Issue l,January(2002), Distributed and Parallel
makes our Distributed Database system more reliable and Database,, pp. S-32
more functional. (9) Scfcroglu, ff., UJu, C.K., Civanlar, M.R., Ke~pf,
In the above figure, a distributed storage system is J. "Congestion State-Based Dynamic FEC Algonthm
invoh·ed a set of cluster storage subsystems. Multiple for Media Friendly Transport Layer",(2009) IEEE PP·
fragments of a file can be stored either in storage nodes 1-10
within a single cluster storage subsystem or in nodes across [I0)Mao, P., Xiao, Y., Qu, G. "Hybrid Traffics Congestion
multiple cluster storage subsystems which consist of a Control Based on 2-D Hurwitz-Schur Stability",
number of storage nodes and a aatcway. Storage nodes are December (2010) Int Conf. Control, Automation,
divided into different type groups, each of which represents Robotics and Vision Singapore,pp.7-10
a level of security wlnerability. (11) Shiang, H-P., and van der Schaar, M. "Content-aware
tcp-fiicndly congestion control for multimedia
7. CONCLUSION tranamissioa" (2011) International Conference on
Acoustics, Speech, and Signal Proccssina
In this paper we studied relevant works done by different (12) Rahman, J., Saha, S., and Hasan, S.F. ,.A New
researchers and introduced advancement of database Congestion Control Algorithm for Dataaram
security as a result of technology changes. We proposed a Congestion Control Protocol (DCCP) Based Real-time
robust congestion control model with fragmentation. This Multimedia Application", (2012) International
technique has applied to the Distributed Database System in Conference on Electrical and Computer Engineering.
the form of clusters. (13) Zhou, D., Song, W., Shi, M. "Good put Improvement
In our model we proposed Clustering of different data- for Multipath TCP by Congestion Window Adaptation
sites "hich aims to truncate expensive adaptation from in Multi-Radio Devices" (2013) IEEE Consumer
Relational 10 Distributed database. lo such database system. Communication and Networking Conference.
security of data considered to be important in cue of
upgradation &om relational database. So, to overcome &om
all the problems we proposed congestion controlled
frasmentalion for relations in heavily loaded centralized
database.

REFERENCES

[I) Ozsu, M.T. and VaJduricz, P. "Principia of Distributed

Database Syatcms" Third Edition (2011) Springer New . Arvind Kumar Maurya is MCA
and UGC-NET qualified and pursuing Ph.D. la the
York Dordrecht Heidclbcra London.
depar1ment of Computer Science DOU Gorakbpur
(2) Huans. Y.F. and Chen, J.H. "frqment Allocatioa in
Distributed Databue Design" No. 17, (2001) Journal University, Oorakhpur (U.P. India) under the supervision or
of Information Science and Enameerins. pp. 491 -506.
Dr. U.N. Tripatbi. The area of rescarc:h iatcrcst ii
(3) Khan, S.I. and Dr. Hoque, AS.M.L,.A New Technique Dilll"IDUled Database Security, Networkina. Arvind Kumar
for Databuc Frqmentation in Distributed S)'lleml" Maurya bu publiabed 5 papen in different national and
international conferences/Journals
hnp://dll.doi.cqll0.5120/940-1311 Volume 5- No.9,
August (2010). lntemationaJ Journal of Computer
Applications.
(4) Hababch. J•..1...,.-ovina Network S)'IICIIII Performance
by Chaltcrin1 Distributed Dacabue Sitca"
hnp:/ldx.doi.ql 0.1007/111227-010-0436-9 Volume
59, Jaue Uanull)'(2012). The Journal of
s ~ a . pp. 249-267.
(5) Ccri, S., Pelapai. G. "Dillribuled Dalabases Principia
and Systcall" Sa:oad Edition (1984) McGraw-Hill, Dr. Surya Pntap Sinab is MCA.
New York UOC-NET qualified and Pb.D. &om the depanment or
(6) Abdalla. H.l."'A N9w Dala Jle.Allocalion Model for Computer. Science DDU Gonkbpur U~vcrsity, Gonkbpur
Dillr11'uled Olllhw $ , - " Volume 5, No. 2 (2012) (U.P. India).. TIie uea or IINll'dl anterat ia D111bw
Security, Netwamq. Dr. Surya Pratap Sinab bM publilbed

~ ·J. J<.,. ~'1~ )II

 •

11 1 h. 011 11 17 7
\ 1' • • 11 ' ' · " ' " •' , , " '· 1111, 111.111011.11 .l,,urn:il nf .\,h l lll"\II kl·,urch In ( ·1111111u1rr ~clrllet. 11 (II), ,rpl CIll 2" ~' ' -.\-'!

more than 30 raren in different national and international

confcrenccslJoumals.

- - . - - • Dr. Upendra Nath Tripathi is

I
Assistant Professor in Department of Computer Science
Dr. Manish Mishra is Assistant DDU Gorakhpur Univmity, Goralthpur (U.P: India). He has
Professor in Department of Electronics DDU Gorakhpur 15 ycan of teaching and research expenence. He_ has
University, Gorakhpur (U.P. India). He has 15 years of published 50 papers in various National and In.t ematlo~I
teaching and research experience. He hu published 55 Journals/conferences. Hi• area of research interest 11
papers in various National and International Journals/ database systems, networking.
conferences. His area of research interest is Computer
Technology, fast processor design.