1

Modle3

N. SQL B2q Pala11anogementlongaDB

nd Cass ando

LovoducHe ohee t cos2sts of

d Ret?bulel Sytem
e

use
B2q Data machees a r e Astaib ekd
data n od es at clusteas
malsple
The tasks
Sojtiaas e Components.
nod e c lust ceg
dota al

shibsetcol-Compul
gArelaect age
Fealuses o
aleiande
ab?|2t ane fanlt
d o Ch cased Hels a clusts,
a
clusks lasl
laslr
to
ofmach#n eg
a aSaqmeot

mac Coe ue Lhe cD AG

Rut a he
hes he data
at numbn of
Ohen he data sets neplicate
toleaanie 2n C e a s 9 ugf
nool ethe ault
to fosrIL eplemet ard olbe
Flextb?\st 2l eeas
K,onmat
sNEces?) a d2st?buutcd eo2
neo

data oto Aefaeatsebs|

stof cpasts of dkacot
By shddi
clusts | Sevess
daanoel es,
o d2shio ceed
foCsease
a
Fo
palrq Po
A Sp ced: The
Cam

AS
Campulng Sqdeo
a
shaiols uo pasa llel on olsddc

to clutas ndep eoleotty

d t a zodti
ofalas.sje elatdoa
shasdeso
57 SealaL:b: B Canetdesrag l9 fo
o
didslad-Fo Camp
eo Mo oshosds
a
enpanols-ecttheg
dalebas
dallae o 4epns 1Dheo h e lo. hasl
machio es oupl
oCAcasrnthe
e addiq mo le
 Pho Ved u belsKoolal scalah:e4
aochmased Compulngpoou ound suno19-th Lhe

e Sarme Madoe epaoVeelt h e

Nlo laonst hms O

Veglical Scalalelhy
6Kesoce shasydsng shas cd h eLo8A a mok
.udun-the Cast
mache n Onc netwozk ascheleewe

17 Tesor, cmance:The Collec2or)

pioccadoy 10tte14
PaoV:dug hsaupalodmooce-han Cofrls zad ospekes a

M
he
he i?buked coer1petEnq
L lssu e to r o a b lechoo Ltrq n a lage netoiking
eokostrud n e

Acddaltonal C oca Iaapa LimerE,

3Seunity s k s o dala aro neioClg

NOSOL Iedu ohu

SaL a a PloJarro danguog e bes ed co .laHonal

deio he dala schena,
alasb to LOh 2
SOL CJeahes dlalakaiea opd PDBI1S, hea RDBM
s Lob ulaz d a l foR ORL halaliora algubra
x Reloliong ae a Set of -lglu, obua named alto:bulez,
Tuqle pu deotslieo uoqeul Kaye called a
Canels daBe hp
Thars cliens co SOL dateboses
exhso?L ACID Paepsla,
ohee f stardstR abmdca h co ncReknayA esola io
ancd duabe let
 ACID PAoP~ 2S@L TAansaceons

mut
P o h e transactons
IArom C : l epesatibos be
a ut
anol 9atpteolthes
meet/ Carslete
Lrclone( 3ol| bak).
hesb
Customea
0etdkaOs an ameunt
X2 t 6 dia)b
opexal ion e a s
Lbe
e bas k s balana
modrehe
the table, nect el
unt
Omo e

LO3 ne amouot aWailable

The ransaciont oLst m a m t e r t h e e t g y

o2Cps seskny
Consredt onol t lo.0 the Conseskocy
plsncple
dHeseoce u depoeel amoust
Th ncut

CLre 0 h dsa0n c
lbank, a c e p u
n a

equa h e lost lealance. AlL 3 d t a n e e d - + o

Cons?lent

tvens oueons he databaie

3-A solal 2e 1he two
senco Laled r n e a c h ottheg
a s d done ksately
must
mus petist coa Compleka
ADuodseLEy:The troun Sa io9

loSgL NoSGL (te Mot oola SSL

A caegohy dola tores es

ASc appiaaeh o larabase 9hesheme

Stmple 9elotionsh2ps dynome Sehemao
oeto shaidtn9
The Pssois 0saL 2 ocke o sundasd3alio
Pp7oeuu
oeuh, photess geq dilfeceulit 9oR Coroplex
ohes.
 Bag Dta NosaL ok Mot cnly aL
a t dots not qui ie pecialLad RDBMs :he choreqe
be a cloas
a haidOake-FoA piecacf Sokoqe coo

oide relaltonol lata sBoeyekns

xNoSeL oue oo9-
use-the
lgThe 9ecohole
Thy cuetextble d a l mc
malleple sche.cmas
aNo S L 2 O feroe actuied data. Brq do

Chaiaclesltc o NoSOL data ctoie as4

- relaleonal daa ore9e
2?t a clas
eem 0 H fexîble da mecde

y
elata ascheleetae patens o
e Noso L Valse pai3s.
valueociksname -

data s s
a iy
Colcemfamdy.
Schema
schema ilke
hae a ed
s Alo SAL oot occessaely Data wto
do not uee he ConCap Tois,
Lable , to d s p l e ncels,
ode Ca be 9ap l I c a t e e
cct coe n

2§Joult-tolsant

data
Henca he
Peatues NasAL T
mo ha othe ACID pop e3l?3.
Bla
od oj hRre pAOpeii
ee Chasa chasze oyos0
CAP holac

fi chaa cles.s.d BASE Ppca ts,

Brg lora NosQL o|leon useand-alone ss ves
macto-clave ord pus#o- pe olBibulico models
 CAP CenstsleryAras abtlaBy anol Taslelion[+AcoiLm

havehe ame Valun

Consiskocy s t a o d s o i olL Copt u
Jaie trcl 2l2onal dbms.

stends ad least ene o p fu ayoslebe

Avaslab? La ty fals.
ve.ce
me
fnaclt ve o
Opasleltons
oCo9e

Sandeopasts
ohrh ae adtve b
Pasleleoobs dB.
Fo desto? bded
may not Coopeote

slardfoi Cool2 nualfos solèns

Pau2lo tolesane Los& node pluse
Loho le aye
ase
of messase

eachoble
not
no de

BhosS CAt utem demen skrates \hoet any d

A ad P kqebe
bero syet Canne t oyLasaote C,
be caie t uses Coospn cy Afaiabelaty and palHon
tolia e
Same dalaat
-|nodee obscg ve -tthe
1 Con4 Pchency
he Same-tEme,
ot
Aecoeve a espon Se
aPAvoa Lab?L
Eacb oyust
u c Ce a

3anl2l!o)--toleaan ce lhe sese Contiouee to pesate

ae
meisag Les o Jil
node.

Paraio toleaan ce Canndt be Oveslooked fp

a chit ag eliab?Lsyen
a d stabte selem.
hus to Case o n y netwo h
heue tilura a chetca ay
betthe toll og
 Oo dd be
e

Dalabase must assOs ard thol ansc@es 0

7
old ol Oong dota AP)
unle 2L gecai ves be
atabase chodld not anccO
a
Jatest CT ofthe
data CoP)

CAP beotem splieoa a oetwoals paslele

.e. ae

4t,te ho?u Cosesleney ond oWeslabiliy

mulualy Enclus tVe Cannot
be

Paslalze
Consfetop
Avoalabe toliante
] IPJ

AP cP
CA

CouchDB Hbel
DBMng
Coss a o s e ongoDB
MysQ
DynaDS
CachedB
PossaL Ros

Sehe)a -les odels:

o a xel Lble
N SQLL dala ocl oeceaisly
dz.stibted o d a
Schema Acluales baseda
02- Nosol DE.
OSgle lasqe data sfora
a

or
x No SOL data model alts 9ela nalfen fo one
mohe AcID p Rop tste ofbe dalabose
Dret biry fo|lo us heohem, O.d 3
o Luost murt Le epreseotfo the appk calta)
 Chakacaeetfcs o schea- les Mod el

/Requa he no AHomta Requi ho_e no

P o hnoige letomi9s S
netadatr- lol, data
ef elatashuelak hoLoody
onued. oe done
Mo neel ata Loaad
6dB3
ERmodeli
l o totftal Modsling encapLsnd
oe
ConéLne mase Lhe
r l es but
PAoced Ca b

h e CDd Ve dl agpam epasegts he chaiaeAZEH

mocdel-to' daletonae. ER cJardi toR
Shema -les
enlthy- elali) Medelin9.
Conoeclons bebren
Relat'o o8 fo a dalabele bala Lbe
Vaitotstables dala.
ty acsdami
sad as
exoAable d ubsect
C a n b e C o o n e c - t e d t
atrble opero

ocalLnPe Ensleti on S
20the
ofeaeo elattone
mcelhemal9 eal
Nos6L data tokes LRe no) -

C g egprhe collel
b sto Aolhec Etotmalim
meladata

data dese2ng aoc

Mladata Zt a s t o
ayobsee obreets
Ot-alLhe talohmalaog obo cpa? cdaz datasl,
anal Lhe t -Eokaq
 Toccastog Flx?b?Lshyto4 Daba aopdalio
haddsltonal data
NoSa L e a schena- leg cchea.
be uetuadand tolloD d a l of hx ed
maynot
data llka
Cansest oddaltonal
h dala efoke
and Loslfeu
do Curnents, bl-, ccbooko poges
chaacteue
stus-the
data stoke p o
ence, NoSaL
1lasepudat
Of PoCALas fexeb?laky to darba.
o-f
be hCAao oLatebcue cos
he n0 alfables to
also pe.oikd.
dded Late bndlng +eero e
storea,
od efo NsaL data
BASE P a
flexable
BASE Piop es~t
ands fe basec avasLab lsty
LEA and eveotualcontseg
stands o sofE Soe

by distibalro shard
Besrc Avas Lalelehy : 2 Loe

ma
data nedes Det hsgkas degpa a
rpicatfon
eve t9 h e Pienco
2t en Su, phoCesk lnq
y of ate
oCons?sleoils
bu achittg he Con sBeny
o

eveottQ

n2clenc The Conse cfenu 9Ahsment )

3y Eveotual
databeus neetinq out Some poinE o tme 9
NaSaL
Data CooVsoLs Neoktally to a Cons ?glent
tome frame pecI?atfo)fo_achiu0,
SHat
haut
hout AcID 9ules Ruike Consteken clL along +he
Corsplelo.
PAOCexs En9
Nole Belo krdbookto4
a9 eample
Schema z not neCe azy èNosQL DB, EPi
Tofoleoatioo stohagp x t LeNeh. Dato can oke oard
rel Dwe Oto ho a ho oldge oho db eoke tos
 5
NoSOL Deta Arch:lectuu PtLeins
eoto
losal olata sto ius bAoadly eego 3z zed
ouchslecual pabfars
1.6q-Valeue tore:
ba t the otmpld way o mpleme a Schema- leg
oala SHok tS use Su-value pod A8
LThe dala rettwal es fost to - ualue paiks
dola sBoRe.
aqe dada
raps to
a
L StmTle stig cal)ed Ky
oBLOB Raste Lagp olgect),
Ky Valeu sto acc esses Ue a
paimaiy ky-bi
acCesingth Values. Hence, tt can be eat?ly Sealed

o Ver ouge data.

2 ? s agim?las o hash Tabl hecsa untqe ky
polos t a cpasltulas em
adal
Valuee Volu
Nalue CAshtal elas: B.E
Sem VI
Banch; a4e,
Cell.lo 1234 56 1890
Woue - ,u Caegocd
e lod ) . E
ot
eel| q0&765 u321

Noeles

umbo od ky Volur
pau can e a Vuy
Jaug unb g.
19.5 ky- Val Pod At eletsol
Pettee
 Ddvaege -value sleu
1 Delastose Can ete/ ony dalaye to 1 bl
19fomatog a adata,Jsese
BLOB.
sto stouRthe a

CHext hfutext :omoges, id uo and Ardio) andutn-6.

9abteved
Soume BLoB
BLOB oheothe darta f

he Values anod 9.ebans-6e Vaue

ya B u Puegcuet
o Singe eem. hese blues con be of any ufe

oloctc
3y y vala sto6es Aeventually CondescE
Oldued
Key Vale chata sto&a n
A Kuy
Ay hauasehd cal
toeh oY

Fiu Value eto ke.

Belwned values on queszu ca be wed o Covwt To7D

Glunns.
l8stsable.olumos,da- eurneddc and
Gyey Qccto qinesa-ec f} eslarble and an be
qSeuuh o
plseteo many fomat s,

aValu
CLoqf cal paltB
names to tmogas ol-e2
aqueel-neepente Call.
E S T - 1Web SuuNl ce call
iYS9L -Gu el:s,

ead a
The l Vale etosePhoV¬d Lehe clseol to
Nalus ucin tol low

i qe Cly) twns the alue a s s o ciatd t 5 Ry

t h e Value sotth he
8 P(ky, Value) > 2 asso ciacs
a Value kuy t
Fuy ane updat
alcd phesent
1y Maltt- gut luy hy. N ) e t wns the Iest
oValeses oue o tialed
 6
1Delee ( 2 tremoves a
kuy aod ets Naluo
h e data to ne

Value efpie Jyheletua Potltso

Snetad to 9s oky.
odeKtode ee ae
Oe
maiotalned o9 Ka. Volues, thus
a ubset oValuee es not Seagchoble.

tt -Value sfoke dos not p3pUede radeltone

dodabae Capouo?LILiee, i AeID pAopeHU

tsansaCL?ons

F?7Lesotcl ofn9he lnfque Value as Kys,e ay

become dilfult hey -he voleum data neas
V)ucAieCansot b paoommec oo 2ndsosdusal Uelus

ompa¬ioo BebOcaDAdi tonal relolformou dat

ne iy-Valu fore
med el
Aad lsonal Rlolional ky- Value stote
Dota 71od el
1Reult et besed e CeueRZ aeteo a

ro Values,
ngle Etem
Values ao04 p No
Duye a l a scs o
7 tndey es en
values
nder eo
aSame data r e Valu
r 37
-Hny data ype Valuts
to
t9 Columm&
 EXample Cencupt Hah k
Decueme h a t heudeot
Deuesto C. Example
an

Cach udeo sade shed

ome
Fa, 6
Oh a . od Cocordo
No. oyalu
ey hos
a

To t. sGPA d CpA valus.

y-Value aiie,
be uwed
0ll h e hash unctroo
t 00 0

Hoeh feuncttog qencuales as Jndu Er fes k

a15 shelo edcally be unfOyue ard chod o opuly a p
to' a

t S a No. 0-+S Peao dtgtts ooly, Copated

he moly huy . used ag

Mo. o heua etcs (0- a55 i
fop o he bah fenetto.

hat tofel 0 . eaiu Cie

paueot
s Sume
bettoesn dot molrces beluoeen00t 99
udeot ame ouy ConcPS o Seveal Chasaeeas
bt toodey Le usl tue duqHs,

o. Dotum ent - lohe

Douunnt sto A petorance

LS calab? lehy VosLe depede aloindl ooens

o.0 o1padD abd, objcot ad
nph -acloi
eatuus unen-sJol

2 stoh the st uetusrd daa

t o 6aqe hae a qea liake ty Rtotel aor
3y p a a Ctoh Ey eelel huaLchies,
A luey tng 24 casy. er x. stnq secioo lo, ub-Sechon No.
 A.
5 No object relat-fonalL capptn9 tnabl es hH edy
Seasch fsom a 2ool o a do Curneot trei

ko_sa ctlon o the documetchohe ehib el -Au

Phop esies

ypecal ses Do umet stot au

1 0u Do Cumeots ety Joveto iy re

i tooms D a ir Do Cum eot Epehange

YDo Cument eaueh.

Dotusnent Arhaece Palleeo and Dsovua tzcsashi

Qtructune

BRANCH BRA4CH

At VE
ALOE NALUe V-LUE

kdy A <]4
ke t</e7

4dy 4 4|4y 6
f6K1
17
bThee Aepiesenla-lToo
Ka mL Douetomen tsgroent
 Doument G||ecHe
A Collect lo odala can be t mO 0ar-fok
do Cuument toke
enanegi2
a
lone
deuunet Gllectteo ase
A w
Th a

StonglaL to a

1 Gaoup
ou the dotuuenerts gef
Qretule f a le yshem
Lcto Ry
detumenl zenashves
Enabls nave9atirqtroeh
doumes anol eRag
egically gouptng Bnt Lay
Rules E l e peamPssto95, tndep
ew
and t9jes
bust ne

3y-Collectto co Cogtei) a6 Collelions olio

3abuas Dadn.

LTcb a dala etoses ue ho0s and Columos

LRoo-heod Pddd gnOy Lsed a
be ky hreh acts a

ond 9ave cnulleple valu9 fo)the Sacce4sve Colurnms,

9 he o0.

L 0 F ( cmeLine TR.ansa cto Tasces es t

Fo-emok o0toamat data

Al piovede Coluno Toienat storeqee

his, tohe
1an- Memoky o. baseol data 7,

Past Column s
e
he Ao mermo oddg.us
Su CCeg2EVe Colums at SuCC C erememo
OdValues to
oddsis. Zt maku oLTP eosi,

Glumo based data z-L bas kup

2 a -emoy SuCLc4°Ve

i9-he Ll Colum) o- eachZod a-tA omemo y -cddguKu

Coluos, N can be a
vetu Loige No.
a c a be upto
a hot ?tmakes OLTP
Obect Deta.store
olject stoe
2Re epotoRy
to o

documens -eldes,
L Object like(Psle amaqeo

bust nes upo 1A:)

LSy
y ele pieVtdes tofeomalto) Dkalenae
ne-da-ta
ilse
Last rmoddsheol onguage _us eel
Caharteoy date
access quimeS$ ons
C C#, Ctt,Pt+ho ano elbes, also

LCueto oefndota phoved es otoomasoo Dskie Sebeel

Coceq hosing pec3.rmtSSfons

b Aapb Database
e Way oplecoot a ata cme f o ue

ghaph database. a'sa higbetble

Alo. of edges cao be
LAnyAlo.of nodes and any
add e d t o Xpano a giaph.

LThe Cemplex ehythtgb a r

Vodgable OtthSelabrlc.
GLapb 0 6 dala oades .cleo
Totes conncceo pio VEedes oge
best dalabcuestem eohe a elatfoo shipe
he

and p hor Celecal valsues

LDocta sloles loas ed c qapb heoRy elalto9

ieG-(E, v)
Oh
E det of edges e,
9 StofVulrces V, V
Nodu epseiernt elsoo objects Fdges co code eotode

realio shap betoees noo) es. f o ome opeialions orl

s t n g gRapb modelk
E2Socioul letooils z Conoeced pople
 Chava cder-ctes a qeapb Dlabase
lse RDE we s4ROL

Iy ttepectali zaol qutid Lanquges, e l e -e dala

daa b
dotabaie etem, ohtch ned
mo d ele e

Ceeles o Ky-Valucs decumets

detume9ts
Campleely ditfeent ny h a
d l a e t o i mod ls
a

Columna and obgect 9o a

det o
bypes- edge
3y
y at
dt
ca
bave hypei- edge
Veefrces a
hyp Aap b
qncaliralto gAapb
bypesq kapb
l.. Ves-tice
10hrch as edge capot ay
S:2e ncecoh.do
amallala
A> Censts ts a Gall ectt beboees gopb
s
wheeb bave Complex 2lesa clfoo
Nacdes.
nedes anol hypesg *opb-

aph DBs Ne04j, l gaegiasb,

olsle Graphs, Tlas and Flodk DB.
reapb
NoSL rcheecl aal Pdlans
6Vaiatsons D Value pairs
a sL-Hable, by
LL Stx
StX da a chtJetuRs ad objet.
graph
o-memoky
Columofame ly, do CumesE,
due o
oceol VaQzalions
axcbglectese omay
LaSele ched Hesce, f4 o be easte
beufne R s eadkements.
Cormpeleire
aschelect ie and lo ea

od voteoge
 NOsaL To lanaqe Biq Dal
Ustog Nps@L to anage By Dela

Limtls-the Suppo st fo o t
eIes,
LNOsGL
L NosGL
spouematix lse Columna-faotly

Lchasa chetecli ce ocsy Crealto and hs

PAe Cesi29 peooL Scalab? Ly and stoveabtlsty
much hiqh mago elde
an
CLffs he Supoal o-AcID qAepertr
L 3t aa
Eoslead uppoks ehP and Ehst piepesi ,

NoSQL olltont B:q-Dala

Btg datr Solsos nceds Sealable eloreqe o TB 87B
stoi data
-6a databese jotos
chepptn9 of Sup poAt
*9eevual
dishebeted SAvelsoqethe es a

Astferty
Cletss.
Coucb DB, Dynamo DB,nqaDB
or

Sol-elo Such a
mahe trans a cttons
CPf heokeco,t
Cols asdia Folleuv

asea a v eas?eg tv Sale.

be Yasleltentq toant
A gsolu-log must
Chaia Cteal sle ce o B r Data NoSOL £ol-ico a

ecalab LSty
y High nd tai
hoZo olally, whet
-t dul gned t o Epand
addL moke mah foe a data ned (SeUL
t ca

otohe peo 0 LRO oesS.

Suy pocl to Rpiato)

Talleple coptu dota ctores aaos
aCao illeple
nodes a clust whee totuiu ht9 aneaabeRsly ad-ox
 betoble :
3y Deste
Dala oluelsons pelmet heshasdsg ard
B:q cna leple chusteas shidh
dastabaeling o haisls o
pettrmande
a rohpud puct
enhanceg

e SAL SaveRs which ae les penskse

47.sages
teols
57 Usoges Opes.SousCe
Schesma- less data Mode
6y uptori to

t o E o t e gtahed
cachTng
uppo losGL
NosoL pBi ae
uolske-he s6LIRDBS.
Sy No Eftexsbelety f
oo shructuiec woy
Axtble (N TTgl) and have
data.
t o o ?oq a r aotpudati9

eLpe oBDrla Paeblems

PRo bleo due to imelaltoos
B:q da-ta pio blemt aigse

oSOL cud otthes DBs

Theelle otnqpes poblem:faceo bj usteg B:g
D t c olal-fo ng

dt neade the Salabla stosage and use of sehibuded

L es clustca
SR VAS tvgetheeg
S o l e t must dsop suppotoa be datakase jotos
ane bas-the qratest ce
DB 8 opeo- Souce
o? NoSGL a

me e a k o e a also, beaus e
the
olo uct at the
ame qrtatett
tancdaids 16 xlase L Aata
her oot masy d e m d
stol.

he
Compagzsio lose 0 SaL RDBNS b
ollo0nq etur
 10
NoSL Doela sOL
Teatus
tore RDBYS

1ModeL Schema-led Relaltooal

Mod el Med el

hey-Value based
dala iehefetui able
Colieno-feamely, beue
dL
Patteans
Docum eot grp)
a cbsect lsoused

3 Scallable HootZontally Scalbble Vestteally

Scalable

As Use SGL No
57 D a t a size Lasge d e t a
Paefers Laige
prefcscoce
daleseis Sets not peksu

G Coos PSlency Vasiab Le ce

Vendo aS post Cpe o n e

3 AcID piopeits lay not Quppoal

Breed, s CP
shiel
e o Hem oue fdlous
BAS pAopeylis

Shaied-Nethinq.chs lee Lre For Bq ale Tad

The Colarons Leooables nelak by a helaioosks .

nelallo algebfafC tgual toy pecafits +be Aelalieo. hy

Shase bet.oas too o maleS0FablG fo RDEMs

Shase Nothi1f a clusta arckaleetuie-A nede dots oct

shas dta oih any -the node
 B:q-Pat stet consfisks of sascieleetue BeData store
easlypauelt gto chasols, pocSo9 Psocesiesthe dafeg
de
quesies data h difeseot u at eah oo

cc
HesCLhe data qseceua eg paLall el
rcependety
Dode'e

he Feat ases ofsN aschefectwre a j

Each d e o e f s nomemoRy cheltnth.s

I Jndeperclence
Comp.talfoma self- SaflFEene
Paocie s a CAatfey

l o s s l e CaLss

selt
Onothe Cok
Healsn:
nodecoctiont og a a shasdd: ach nede t o r
37
37Each
a shasd

A No Nezooiko Conoech