Description
Business Intelligence has reached a new level of importance for decision makers.
Busi ness I nt el l i genc e
i n a Real - Ti me Wor l d
Whi t e Paper
Conf i dent i al
1.866.Net ezza www.net ezza.c om
Al l ri ght s reserved. © 2004 Net ezza Corporat i on.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 1
Abst r ac t
Busi ness Int el l i gence has reached a new l evel of i mport ance f or deci si on makers. Capt uri ng
dat a— f rom cust omers, part ners, empl oyees and operat i ons— has al ways been a part of runni ng
a busi ness, but usi ng t hi s i nf ormat i on proact i vel y has become essent i al . Busi ness Int el l i gence (BI)
can be t he di f f erence bet ween react i ng t o probl ems and ant i ci pat i ng opport uni t i es, t he di f f erence
bet ween educat ed guesses and proact i ve deci si ons based on f i rm f act s.
Si mpl y st ori ng vast amount s of dat a does not ensure ef f ect i ve BI. To be usef ul , BI must t urn
dynami c, det ai l ed dat a i nt o i nf ormat i on, and make i t avai l abl e i n real - t i me, at t he poi nt of a
deci si on. It must be accessi bl e on- demand— when and where i t ’s needed.
Yet t here are chal l enges t o real - t i me del i very of BI. The amount of dat a i s growi ng exponent i al l y.
The number of BI users i s const ant l y i ncreasi ng. Exi st i ng dat a warehouses have been bui l t upon a
pat chwork of hardware, sof t ware and st orage t echnol ogi es t hat can’ t keep up wi t h t he growt h of
dat a or t he i ncreasi ng user demand f or i nf ormat i on. As t he st ress on t he syst em grows, query
response t i mes move f rom mi nut es t o hours t o days. As a resul t , managers are f aced wi t h t he
choi ce of ei t her post poni ng deci si ons or act i ng i n t he dark.
What ’s needed i s a new BI pl at f orm desi gned t o provi de real - t i me access t o Busi ness
Int el l i gence at t he t erabyt e l evel . Thi s new BI pl at f orm, an i nt egrat ed sol ut i on of
hardware, sof t ware and st orage, i ncorporat es t he best f eat ures of dat a
war ehouses and adds r eal - t i me avai l abi l i t y and r el i abi l i t y. Onl y
an i nt egrat ed sol ut i on can del i ver opt i mal perf ormance f or dat a l oadi ng,
ad hoc queri es and compl ex anal yt i cs. Wi t h such a pl at f orm, t he promi se of
BI becomes real i t y.
The Importance of Business Intelligence
Busi ness condi t i ons are i n a const ant st at e of f l ux. Sal es pat t erns change f rom pl ace t o pl ace and
f rom t i me t o t i me. Currency val uat i ons shi f t and al t er prof i t margi ns. Suppl i ers change t hei r
del i very schedul es and t hei r pri ces. Cust omers become more educat ed and t heref ore more
demandi ng. Bal anci ng on t hi s shi f t i ng t errai n, busi ness managers are expect ed t o del i ver st eady
earni ngs growt h. Somehow, t hey must smoot h out t he bumps and ant i ci pat e t he changes.
The promi se of BI i s t o enabl e organi zat i ons t o underst and change, t o i dent i f y causal f act ors
t hrough anal ysi s of dat a by regi on, currency, cust omer or ot her rel evant di mensi ons. As t he CIO of
a l arge chemi cal company sai d recent l y, “ If you rel y on averages, i t ’s easy t o be mi sl ed, but i f you
sl i ce t hrough t he dat a, i t ’s possi bl e t o see exact l y what ’s af f ect i ng cost s.”
The di f f erence bet ween Busi ness Int el l i gence and busi ness i nert i a i s t he di f f erence bet ween
Wal Mart and K- Mart . One company uses i t s dat a warehouse t o rel ent l essl y hone i t s operat i ons, t he
ot her f i l es f or Chapt er 11. Wal Mart capt ures poi nt - of - sal e t ransact i ons f rom over 2,500 st ores i n
si x count ri es i n i t s t era- scal e dat a warehouse. It uses t hi s i nf ormat i on t o ask quest i ons l i ke
What ’ s needed i s a new BI
pl at f or m desi gned t o pr ovi de r eal - t i me
ac c ess t o Busi ness I nt el l i genc e at
t he t er abyt e l evel .
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 2
“ For whi ch st ores and f or whi ch mont hs was Fl ex Shampoo i n hi gh demand but short suppl y?” Wi t h
t he answer, i t can opt i mi ze i nvent ory and capi t al i ze on pri ci ng opport uni t i es t hat exi st t ransi ent l y
at a l ocal st ore.
The i mport ance of BI i s not l i mi t ed t o ret ai l ers. Tel ecommuni cat i ons carri ers use BI t o i dent i f y and
mi t i gat e f raud. A del i very servi ce mi ght predi ct whi ch vehi cl es are most l i kel y t o break down and
where. A bank mi ght use BI t o i dent i f y cust omers who, based on t hei r recent act i vi t y, are l i kel y t o
t ransf er t hei r account t o anot her i nst i t ut i on. The possi bi l i t i es are l i mi t l ess.
Busi ness Int el l i gence makes ent erpri se dat a act i onabl e. It uncovers t rends and pat t erns t hat
mi ght ot herwi se go undet ect ed. Managi ng a busi ness on i nt ui t i on, educat ed guesses or averages
i sn’ t good enough anymore. To be successf ul , a company needs a f oundat i on of accurat e, current
and compl et e i nf ormat i on.
Pr obl ems wi t h t he Cur r ent BI I nf r ast r uc t ur e
The Vi ce Presi dent of Cust omer Knowl edge Management at a mul t i - nat i onal consumer goods
company i s f aced wi t h a seri ous chal l enge: managi ng t erabyt es of cust omer dat a t hat doubl es
every year, al ong wi t h ad hoc queri es and schedul ed report s t hat grow ever more compl ex. The
company’s market i ng and f i nanci al anal yst s need t hi s i nf ormat i on, and are pl aci ng heavy
demands on t he company’s BI i nf rast ruct ure. They need qui ck responses, and t he exi st i ng
i nf rast ruct ure i sn’ t keepi ng pace. Unabl e t o obt ai n t he i nf ormat i on t hey need when t hey need i t ,
t hey must choose bet ween del ayi ng deci si ons or maki ng deci si ons on i ncompl et e and out dat ed
snapshot s of i nf ormat i on.
The probl em f or t he Vi ce Presi dent i s t hat merel y gat heri ng demographi c and operat i onal dat a i n
a warehouse does not ensure ef f ect i ve i nt el l i gence. To be usef ul , she must make t hi s dat a
avai l abl e i n real - t i me f or BI, at t he poi nt of a deci si on. It must be accessi bl e t o t he company’s
hundreds of busi ness users around t he worl d– exact l y when and where t hey need i t .
The company’s BI i nf rast ruct ure i s great l y chal l enged by t he demands of st ori ng and anal yzi ng
i nf ormat i on. It i s a pat chwork of t echnol ogy t hat has accumul at ed f or a vari et y of reasons over a
l ong peri od: many f l avors of general - purpose DBMS sof t ware, a part i al l y i mpl ement ed mi ddl eware
st rat egy or t wo, a col l ect i on of di f f erent mi d- t i er SMP servers, assort ed di sk arrays and myri ad
end- user appl i cat i ons rel yi ng on di f f erent dat abase and communi cat i on st andards. And t hi s
pat chwork of t echnol ogy i s hel d t oget her by expensi ve dat abase and syst em admi ni st rat ors who
are st ruggl i ng t o keep up wi t h user demands.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 3
As more i nf ormat i on i s added t o exi st i ng syst ems, t hey are sl owi ng dramat i cal l y and becomi ng
unrel i abl e. Onl i ne i ni t i at i ves, such as capt uri ng and anal yzi ng cl i ckst ream dat a, t hreat en t o
overwhel m t he i nf rast ruct ure ent i rel y. As BI demands and cust omer dat a grow, t he Vi ce Presi dent
of Cust omer Knowl edge Management and t he CIO are f aced wi t h hi ri ng expensi ve DBAs and
syst ems admi ni st rat ors and maki ng mul t i - mi l l i on- dol l ar i nvest ment s i n hardware, sof t ware,
net worki ng and st orage, i n a l osi ng bat t l e t o mai nt ai n current l evel s of perf ormance.
Thi s si t uat i on i s a common one f or Gl obal 1000 compani es across many i ndust ri es. Over t he past
decade, hi gh- prof i l e management i ni t i at i ves have st rai ned t he current i nf rast ruct ure wi t h t hei r
need f or near- i nst ant access t o dat a across t he ent erpri se:
• Cust omer Rel at i onshi p Management (CRM) i nt egrat es al l areas of busi ness t hat af f ect a
cust omer and requi res det ai l ed, comprehensi ve i nf ormat i on about cust omers across al l t ouch
poi nt s. Thi s i nf ormat i on i s requi red f or coordi nat ed market i ng, sal es and servi ce act i vi t i es.
Ti mel y dat a i s cruci al i f a company i s t o communi cat e consi st ent l y wi t h i t s cust omers.
• Operat i ons management al l ows compani es t o i dent i f y t rends and pat t erns and st reaml i ne
t hei r net work and f i nanci al operat i ons.
• Suppl y Chai n Management (SCM) enabl es a group of compani es t o work t oget her t o opt i mi ze
product assembl y, del i very and i nvent ory. In t he past , manuf act urers cont rol l ed t he pace at
whi ch t hey bui l t and di st ri but ed t hei r product s. Now cust omer orders dri ve t he process and
manuf act urers keep pace wi t h cust omers’ needs f or product f eat ures, order f ul f i l l ment and
f ast del i very.
• Ent erpri se Resource Pl anni ng (ERP) coordi nat es i nf ormat i on f rom depart ment s across t he
ent erpri se, i ncl udi ng manuf act uri ng, order ent ry, account s recei vabl e and payabl e,
purchasi ng, warehousi ng, t ransport at i on and human resources. Compani es rel y on t hei r ERP
syst ems t o hel p t hem anal yze f i nanci al model s, cal cul at e i nvent ory val uat i ons, perf orm
f orecast i ng and t ri gger repl eni shment cycl es.
• Part ner Rel at i onshi p Management (PRM) al l ows compani es t o communi cat e and i nt eract wi t h
t hei r resel l ers and part ners. As wi t h CRM, t i mel y and personal i zed i nf ormat i on l eads t o more
ef f ect i ve and prof i t abl e part nershi ps.
Al l t hese i ni t i at i ves share a need f or more t i mel y, det ai l ed and current dat a f rom a broad range of
sources. They have al l resul t ed i n l ong, cost l y i mpl ement at i ons of sophi st i cat ed t ool s and
appl i cat i ons t o al l ow easi er access and cont rol . Most of t hem requi re bot h ad hoc queri es and
compl ex anal yses, such as dat a mi ni ng, whi ch enabl e compani es t o uncover pat t erns and
rel at i onshi p by di ggi ng t hrough massi ve amount s of dat a.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 4
Wi t h al l t hese st resses on t he current BI i nf rast ruct ure, i t i s not surpri si ng t hat many busi ness and
t echni cal pai n poi nt s have emerged:
• Sl ow access t o i nf ormat i on i nhi bi t s spont aneous queri es, whi ch resul t s i n l ost opport uni t i es
• Compl ex queri es and report s requi re hours or even days t o process
• To speed processi ng, dat a must be sampl ed and summari zed, l i mi t i ng dept h of anal ysi s
• Gat heri ng usef ul i nf ormat i on requi res t rai ni ng. Yet t he peopl e who need t he dat a are t ypi cal l y
not t he same peopl e who can run t he queri es. Thi s creat es bot t l enecks i n t he organi zat i on, and
resul t s i n l ost opport uni t i es.
• Produci ng predi ct i ve model s i s di f f i cul t and of t en requi res a st af f of st at i st i ci ans
• Scal abi l i t y suf f ers as dat abases grow beyond hundreds of gi gabyt es. Cost s i ncrease
unpredi ct abl y and di scont i nuousl y, yet perf ormance remai ns poor.
• Dat a l oadi ng i s sl ow and pai nf ul and requi res a medl ey of ETL t ool s; queri es run on out dat ed
i nf ormat i on can yi el d mi sl eadi ng or i naccurat e resul t s
• Mergi ng or st andardi zi ng dat a i s t i me- consumi ng and di f f i cul t because dat a i s hel d i n a
vari et y of f ormat s and on l egacy syst ems
• Parsi ng of dat a mart s f rom l arge dat a warehouses t o ret ri eve dat a i s di f f i cul t t o achi eve i n an
accept abl e t i me f rame
• Teams of dat abase and syst ems admi ni st rat ors are requi red t o t une syst ems and queri es t o
achi eve accept abl e perf ormance
Mai nf r ame
Dat a
ERP
Appl i c at i ons
CRM
Appl i c at i ons
OLTP
Oper at i onal
Dat a St or age
M
e
t
a
D
a
t
a
M
a
n
a
g
e
m
e
n
t
O
L
A
P
S
e
r
v
e
r
R
e
p
o
r
t
i
n
g
a
n
d
Q
u
e
r
y
F
u
n
c
t
i
o
n
s
D
a
t
a
E
x
t
r
a
c
t
i
o
n
,
T
r
a
n
s
f
o
r
m
a
t
i
o
n
,
a
n
d
L
o
a
d
i
n
g
Dat a
War ehouse
Cust omer
Pr of i t abi l i t y
Fi nanc i al
Pl anni ng
Busi ness
Devel opment
Mar k et i ng Ful f i l l ment I nvent or y
Dat a Mar t s
Mul t i pl e sour c es c ont r i but e t o
i nf r ast r uc t ur e st r esses.
Busi nesses must c onst ant l y
j uggl e new dat a wi t h ar c hi ved
dat a, new sour c es wi t h exi st i ng.
Conf i dent i al
Al t hough BI i s wi del y recogni zed as i mport ant , t oo many i ni t i at i ves have ended as cost l y f ai l ures.
The expl anat i ons somet i mes ment i on ‘ i l l def i ned goal s’ , or ‘ i nsuf f i ci ent commi t ment f rom
management ’ . But of t en, ‘ i nsuf f i ci ent commi t ment ’ means t hat t he i nf rast ruct ure proved
unexpect edl y expensi ve. And f requent l y, ‘ i l l - def i ned goal s’ are goal s t hat were promi sed by a sal es
person, but t hat coul dn’ t be achi eved on t he act ual pl at f orm del i vered. The f undament al probl em
i s of t en t hat t he underl yi ng dat a warehouse i s t oo hard t o set up and admi ni st er, t oo hard t o use,
has perf ormance t hat i s t oo sl ow and i s unabl e t o keep pace wi t h growt h.
The Hi st or y of t he BI I nf r ast r uc t ur e
How di d t hi s exi st i ng BI i nf rast ruct ure evol ve, and why has i t been unabl e t o keep pace? Decades
ago, t he ori gi nal BI i nf rast ruct ure f ocused on dat a warehouses t o archi ve dat a f rom operat i onal
syst ems. The purpose of t he ori gi nal archi ve was t o st ore l arge amount s of dat a f or some
subsequent use, not t o anal yze dat a qui ckl y. Anal ysi s was consi dered l at er.
The ori gi nal i nf rast ruct ure di d not ant i ci pat e t he ki nds of needs t hat woul d ari se 10 t o 15 years
af t er i t s bi rt h. Generi c hardware, i n t he f orm of st orage medi a and f i l e servers, was added as
needed wi t hout seri ous t hought of f ut ure requi rement s and demands. Sof t ware f or servi ng t he
dat a was creat ed based on i mmedi at e needs. The resul t i ng mi x of hardware and sof t ware
creat ed a makeshi f t sol ut i on t hat users cont i nual l y ‘ t weaked’ t o accompl i sh more.
Bef ore 1990, much dat abase dat a was kept on l egacy syst ems, many of whi ch
were mai nf rames runni ng such programs as DB2. Programmers pai nst aki ngl y
wrot e programs t o ext ract i nf ormat i on f rom t he dat a st ored on t hese syst ems.
The i nf rast ruct ure began t o evol ve spont aneousl y, wi t h l i t t l e hi gh- l evel pl anni ng or
consi derat i on f or f ut ure needs.
Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d
The i nf r ast r uc t ur e has been
evol vi ng spont aneousl y, wi t h
l i t t l e hi gh- l evel pl anni ng or
c onsi der at i on f or f ut ur e needs.
Client/Server Era
DBMS
File Servers:
Undifferentiated
Blocks of Binary Data
General Purpose Boxes:
Scientific/Media/Data/Email/Web...
DBMS:
I BM
I nformix
Microsoft
Sybase
Teradata
Oracle
Storage:
EMC
Hitachi
Veritas
Servers:
HP
I BM
Sun
Teradata
End User BI
and Vertical
Applications
The exi st i ng i nf r ast r uc t ur e t r i es t o
mak e do wi t h a pat c hwor k of
har dwar e, sof t war e and st or age,
some of whi c h was never meant t o
wor k t oget her seaml essl y.
5
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 7
In t he 1990s, personal comput ers gave advanced users access t o t he l egacy syst ems and enabl ed
t hem t o perf orm anal yses. Busi ness users f ound t hat t he dat a t hey coul d now access was
f ragment ed, and came i n many f ormat s across dat abases, f i l es and spreadsheet s.
Dat a use was di vi ded i nt o t wo pri mary act i vi t i es: sl i ci ng and anal yzi ng dat a at one l evel , and
dri l l i ng down i nt o dat a. Response t i me was a seri ous i ssue f or a dat a warehouse of any
si gni f i cant si ze. Throughout t he 1990s, comput er hardware and sof t ware dropped i n pri ce and
i ncreased si gni f i cant l y i n perf ormance. Anal ysi s of l arge amount s of dat a became more f easi bl e
and desi rabl e. Cl i ent / server comput i ng t ook hol d, al ong wi t h report i ng and anal ysi s t ool s,
al l owi ng PC users t o access dat a warehouses more easi l y and f rom di f f erent geographi cal
l ocat i ons. Response t i me and st abi l i t y were st i l l i ssues. By 2000, t he BI i nf rast ruct ure t hat had
evol ved— a poorl y bui l t f oundat i on— was now t asked wi t h support i ng a compl ex, growi ng set of
t ool s, appl i cat i ons, users, processes and demands.
Recent l y, a number of f orces have come i nt o pl ay whi ch l ead t owards a new BI pl at f orm, one t hat
addresses t he probl ems of t he ol d i nf rast ruct ure. The f i rst i s t he wi despread use of open, i ndust ry
st andard i nt erf aces across pl at f orms, i ncl udi ng ODBC, JDBC and ot hers. The second i s t he
rapi dl y decl i ni ng cost and i ncreasi ng perf ormance of di sk, processi ng, net work and st orage
component s— f or exampl e, t he wi del y known Moore’s Law t hat st at es t hat dat a densi t y per square
i nch on i nt egrat ed ci rcui t s i s doubl i ng approxi mat el y every 18 mont hs. The t hi rd i s t he
exponent i al growt h of dat a, est i mat ed by Greg’s Law t o doubl e on average every ni ne mont hs.
Fi nal l y, t here i s t he emergence and adopt i on of speci al - purpose appl i ances, i ncl udi ng net work
appl i ances, medi a servers and ot hers. Toget her, t hese f orces open a door t o a pl at f orm t hat of f ers
revol ut i onary pri ce/ perf ormance f or t era- scal e BI.
Towar ds A New BI Pl at f or m
Today, a new BI pl at f orm i s needed t o rat i onal i ze and revi t al i ze t he exi st i ng i nf rast ruct ure. Thi s new
pl at f orm provi des a f oundat i on f or t he exponent i al dat a growt h t hat i s common i n t oday’s
busi ness envi ronment . It combi nes massi vel y paral l el hardware, sof t ware and st orage and i s
di rect l y f ocused on provi di ng opt i mal response t i mes and scal abi l i t y at t he t erabyt e l evel . The new
BI pl at f orm el i mi nat es yest erday’s pat chwork of hardware, sof t ware and st orage, and enabl es
opt i mi zed access t o i nf ormat i on. As busi ness and t echni cal demands cont i nue t o grow
and change i n t he new cent ury, t he new BI pl at f orm i s desi gned t o scal e wi t h dat a
si ze, scope and perf ormance needs. And i t does al l t hi s at an af f ordabl e and
predi ct abl e pri ce.
The new pl at f orm of f ers an open archi t ect ure, massi vel y paral l el processi ng and
scal abi l i t y and al l ows growt h wi t hout compl ex upgrades or pai nf ul i nt egrat i ons of
new syst ems.
The new BI pl at f or m el i mi nat es
yest er day’ s pat c hwor k of har dwar e,
sof t war e and st or age, and enabl es
opt i mi zed ac c ess t o i nf or mat i on.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 7
As dat a grows t o hundreds of t erabyt es, a company can seaml essl y add t he hardware
necessary f or addi t i onal i nf ormat i on. Dat a i s mi rrored f or prot ect i on agai nst l oss, and t here i s hi gh
avai l abi l i t y t hrough redundancy. Rat her t han over- usi ng one huge dat abase, t he company can
have a speci f i cal l y engi neered, t era- scal e dat a warehouse. Al l dat a can be l i nked and managed
f rom one l ocat i on. Thi s saves admi ni st rat i ve resources and reduces cost s t hrough bet t er
deci si on- maki ng and t he need f or f ewer syst ems and dat abase admi ni st rat ors.
For bot h busi ness and t echni cal users, t here are many benef i t s of t he new BI pl at f orm. Si nce
l egacy syst ems and exi st i ng BI t ool s i nt egrat e easi l y wi t h t he new pl at f orm, busi ness users can
perf orm al l t hei r current anal yses wi t h f ami l i ar i nt erf aces— but i n a f ract i on of t he t i me, and wi t h
t he opport uni t y t o add new BI act i vi t i es t hat previ ousl y woul d have been i mpossi bl e. The new
pl at f orm al l ows a qui cker response t o cust omer, part ner and empl oyee needs. It ensures
perf ormance, rel i abi l i t y and scal abi l i t y, and reduces t he st rai n and demands on corporat e IT.
There are as many benef i t s of t he new BI pl at f orm as t here are di verse busi nesses. Wi t h real - t i me
access t o vast st ores of BI:
• Tel ecommuni cat i ons carri ers can anal yze t erabyt es of cal l det ai l records (CDRs) i n real - t i me
t o predi ct prof i t abl e cust omers at ri sk t o churn and t ake prevent at i ve act i on, savi ng bi l l i ons
of dol l ars
• Ret ai l ers can run comprehensi ve CRM anal yses t o det ermi ne whi ch on- t he- spot deal s
wi l l boost revenues
• Fi nanci al servi ces i nst i t ut i ons can run f ast , comprehensi ve credi t ri sk anal yses f or l oan
approval s, savi ng mi l l i ons i n l oan def aul t s
• E- busi nesses can det ermi ne t he short est cl i cks- t o- purchase pat h t hrough
cl i ckst ream anal yses
• Out sourced anal yt i c servi ce provi ders can real i ze a compet i t i ve advant age by of f eri ng f ast er
response t i mes i n t hei r SLAs
• Credi t card compani es can det ect and handl e f raud as soon as i t happens by i mmedi at el y
pi cki ng up i nconsi st enci es i n buyi ng habi t s or geography. They can uncover new f raud
pat t erns and qui ckl y appl y rul es t o el i mi nat e t hem.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 8
The Net ezza Syst em
The Net ezza Perf ormance Server™ 8000 seri es i s t he real i zat i on of t he new BI pl at f orm. The
Net ezza Perf ormance Server (NPS™ ) syst em i s an ent erpri se- cl ass dat a warehouse appl i ance t hat
del i vers breakt hrough perf ormance and ease- of - use at a f ract i on of t he cost of t radi t onal dat a
warehouses. The NPS appl i ance of f ers:
• 10 t o 50 t i mes t he per f or manc e of exi st i ng dat a war ehouse syst ems at hal f t he pr i c e—
f i nal l y an af f ordabl e sol ut i on t o busi ness users’ perf ormance needs
• Ti ght i nt egrat i on of st orage, server and dat abase component s, and i t s massi vel y paral l el
archi t ect ure provi des rapi d, opt i mi zed execut i on of i nt eract i ve queri es, as wel l as compl ex
report s and anal yses
• Ext reme ease- of - use and rel i abi l i t y— a part - t i me DBA i s al l t hat ’s requi red
t o manage t he syst em.
• Real - t i me dat a l oadi ng, sel f - t uni ng dat a st orage and hi gh t hroughput on l arge,
compl ex workl oads
• St rai ght f orward i nt egrat i on, t hrough i t s open APIs, wi t h al l maj or el ement s of t he BI
envi ronment : BI appl i cat i ons and t ool s, ETL and EAI t ool s, l egacy syst ems and dat a sources.
The Net ezza Per f or manc e Ser ver
appl i anc e i s t he c or e of f ast ,
c ompr ehensi ve BI anal yses,
r egar dl ess of dat a si ze.
Other Data Sources
Operational
Business
Systems
Programmers
Designers
Administrators
Data Mining
Business Analysis & Marketing
I mplementation
Decisions
Business Partners
Ecommerce
Vendors
Customers
Netezza
Performance
Server
System
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 9
The NPS syst em opt i mi zes BI queri es f or maxi mum perf ormance. The archi t ect ure i s desi gned f or
speed and i s l i mi t ed onl y by di sk I/ O. The NPS syst em i s bui l t f or qui ck depl oyment , si mpl e
admi ni st rat i on and nearl y i nf i ni t e scal i ng— i nt o t he hundreds of t erabyt es. The t ot al cost
of ownershi p (TCO) of Net ezza’s perf ormance i s a f ract i on of t he cost of sol ut i ons
current l y on t he market .
The dramat i c perf ormance of f ered by t he NPS syst em enabl es real - t i me
f orecast s and anal ysi s. ‘ What i f ’ scenari os can be perf ormed by t he peopl e who
use t he dat a, rat her t han havi ng t o wai t i n a queue f or IT support . Queri es t hat
used t o t ake hours are reduced t o a f ew seconds, so t hat compani es can respond
proact i vel y t o opport uni t i es i nst ead of react i ng t o probl ems. The bot t om l i ne:
easi er and f ast er access t o i nf ormat i on hel ps busi ness users make bet t er deci si ons.
Conc l usi on
Ti mel y access t o BI has never been more i mport ant . Sophi st i cat ed use of BI can hel p organi zat i ons
reduce cost s, pl an i nvent ory, i dent i f y sal es and market i ng opport uni t i es, opt i mi ze pri ci ng and,
ul t i mat el y, i ncrease revenues. For many compani es, aggressi ve and real - t i me BI act i vi t i es can
make t he di f f erence bet ween wi nni ng and l osi ng t o t he compet i t i on.
Wi t h so much at st ake, ent erpri ses can no l onger af f ord t o i mprovi se sol ut i ons. A pat chwork
syst em of hi gh- end hardware, general - purpose dat abase sof t ware and st orage may cost a l ot of
money, but i t cannot consi st ent l y del i ver real - t i me perf ormance f or rapi dl y growi ng, t era- scal e
dat a st ores. In order t o achi eve t he promi se of BI, a new pl at f orm i s requi red.
The NPS syst em i s desi gned f or a si ngl e purpose: t o del i ver t era- scal e BI 10 t o 50 t i mes f ast er t han
exi st i ng dat a warehouse sol ut i ons, at hal f t he cost . Wi t h t he NPS syst em’s unbeat abl e
pri ce/ perf ormance, l ow TCO and ease- of - use, compani es across a broad range of i ndust ri es are
f i nal l y abl e t o l everage t hei r Busi ness Int el l i gence and real i ze a compet i t i ve advant age.
WP.BIRTW.1.04
Net ezza Cor por at i on 200 Crossi ng Boul evard, Frami ngham, Massachuset t s 01702- 4480
t el 508.665.6800 f ax 508.665.6811 www.net ezza.c om
© Net ezza 2004. Al l ri ght s reserved. Net ezza, t he Net ezza l ogo, NPS, Perf ormance Server, Net ezza Perf ormance Server, Int el l i gent Query St reami ng,
Asymmet ri c Massi vel y Paral l el Processi ng and AMPP are t rademarks or regi st ered t rademarks of Net ezza Corporat i on i n t he U.S. and/ or ot her count ri es.
Al l ot her company, brand and product names cont ai ned herei n may be t rademarks or regi st ered t rademarks of t hei r respect i ve hol ders.
The Net ezza Per f or mance Ser ver
syst em i s t he f i r st ent er pr i se- c l ass
dat a war ehouse appl i anc e opt i mi zed
f or BI on t ens, even hundr eds,
of t er abyt es of dat a i n t oday’s
r eal - t i me busi ness wor l d.
doc_570265564.pdf
Business Intelligence has reached a new level of importance for decision makers.
Busi ness I nt el l i genc e
i n a Real - Ti me Wor l d
Whi t e Paper
Conf i dent i al
1.866.Net ezza www.net ezza.c om
Al l ri ght s reserved. © 2004 Net ezza Corporat i on.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 1
Abst r ac t
Busi ness Int el l i gence has reached a new l evel of i mport ance f or deci si on makers. Capt uri ng
dat a— f rom cust omers, part ners, empl oyees and operat i ons— has al ways been a part of runni ng
a busi ness, but usi ng t hi s i nf ormat i on proact i vel y has become essent i al . Busi ness Int el l i gence (BI)
can be t he di f f erence bet ween react i ng t o probl ems and ant i ci pat i ng opport uni t i es, t he di f f erence
bet ween educat ed guesses and proact i ve deci si ons based on f i rm f act s.
Si mpl y st ori ng vast amount s of dat a does not ensure ef f ect i ve BI. To be usef ul , BI must t urn
dynami c, det ai l ed dat a i nt o i nf ormat i on, and make i t avai l abl e i n real - t i me, at t he poi nt of a
deci si on. It must be accessi bl e on- demand— when and where i t ’s needed.
Yet t here are chal l enges t o real - t i me del i very of BI. The amount of dat a i s growi ng exponent i al l y.
The number of BI users i s const ant l y i ncreasi ng. Exi st i ng dat a warehouses have been bui l t upon a
pat chwork of hardware, sof t ware and st orage t echnol ogi es t hat can’ t keep up wi t h t he growt h of
dat a or t he i ncreasi ng user demand f or i nf ormat i on. As t he st ress on t he syst em grows, query
response t i mes move f rom mi nut es t o hours t o days. As a resul t , managers are f aced wi t h t he
choi ce of ei t her post poni ng deci si ons or act i ng i n t he dark.
What ’s needed i s a new BI pl at f orm desi gned t o provi de real - t i me access t o Busi ness
Int el l i gence at t he t erabyt e l evel . Thi s new BI pl at f orm, an i nt egrat ed sol ut i on of
hardware, sof t ware and st orage, i ncorporat es t he best f eat ures of dat a
war ehouses and adds r eal - t i me avai l abi l i t y and r el i abi l i t y. Onl y
an i nt egrat ed sol ut i on can del i ver opt i mal perf ormance f or dat a l oadi ng,
ad hoc queri es and compl ex anal yt i cs. Wi t h such a pl at f orm, t he promi se of
BI becomes real i t y.
The Importance of Business Intelligence
Busi ness condi t i ons are i n a const ant st at e of f l ux. Sal es pat t erns change f rom pl ace t o pl ace and
f rom t i me t o t i me. Currency val uat i ons shi f t and al t er prof i t margi ns. Suppl i ers change t hei r
del i very schedul es and t hei r pri ces. Cust omers become more educat ed and t heref ore more
demandi ng. Bal anci ng on t hi s shi f t i ng t errai n, busi ness managers are expect ed t o del i ver st eady
earni ngs growt h. Somehow, t hey must smoot h out t he bumps and ant i ci pat e t he changes.
The promi se of BI i s t o enabl e organi zat i ons t o underst and change, t o i dent i f y causal f act ors
t hrough anal ysi s of dat a by regi on, currency, cust omer or ot her rel evant di mensi ons. As t he CIO of
a l arge chemi cal company sai d recent l y, “ If you rel y on averages, i t ’s easy t o be mi sl ed, but i f you
sl i ce t hrough t he dat a, i t ’s possi bl e t o see exact l y what ’s af f ect i ng cost s.”
The di f f erence bet ween Busi ness Int el l i gence and busi ness i nert i a i s t he di f f erence bet ween
Wal Mart and K- Mart . One company uses i t s dat a warehouse t o rel ent l essl y hone i t s operat i ons, t he
ot her f i l es f or Chapt er 11. Wal Mart capt ures poi nt - of - sal e t ransact i ons f rom over 2,500 st ores i n
si x count ri es i n i t s t era- scal e dat a warehouse. It uses t hi s i nf ormat i on t o ask quest i ons l i ke
What ’ s needed i s a new BI
pl at f or m desi gned t o pr ovi de r eal - t i me
ac c ess t o Busi ness I nt el l i genc e at
t he t er abyt e l evel .
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 2
“ For whi ch st ores and f or whi ch mont hs was Fl ex Shampoo i n hi gh demand but short suppl y?” Wi t h
t he answer, i t can opt i mi ze i nvent ory and capi t al i ze on pri ci ng opport uni t i es t hat exi st t ransi ent l y
at a l ocal st ore.
The i mport ance of BI i s not l i mi t ed t o ret ai l ers. Tel ecommuni cat i ons carri ers use BI t o i dent i f y and
mi t i gat e f raud. A del i very servi ce mi ght predi ct whi ch vehi cl es are most l i kel y t o break down and
where. A bank mi ght use BI t o i dent i f y cust omers who, based on t hei r recent act i vi t y, are l i kel y t o
t ransf er t hei r account t o anot her i nst i t ut i on. The possi bi l i t i es are l i mi t l ess.
Busi ness Int el l i gence makes ent erpri se dat a act i onabl e. It uncovers t rends and pat t erns t hat
mi ght ot herwi se go undet ect ed. Managi ng a busi ness on i nt ui t i on, educat ed guesses or averages
i sn’ t good enough anymore. To be successf ul , a company needs a f oundat i on of accurat e, current
and compl et e i nf ormat i on.
Pr obl ems wi t h t he Cur r ent BI I nf r ast r uc t ur e
The Vi ce Presi dent of Cust omer Knowl edge Management at a mul t i - nat i onal consumer goods
company i s f aced wi t h a seri ous chal l enge: managi ng t erabyt es of cust omer dat a t hat doubl es
every year, al ong wi t h ad hoc queri es and schedul ed report s t hat grow ever more compl ex. The
company’s market i ng and f i nanci al anal yst s need t hi s i nf ormat i on, and are pl aci ng heavy
demands on t he company’s BI i nf rast ruct ure. They need qui ck responses, and t he exi st i ng
i nf rast ruct ure i sn’ t keepi ng pace. Unabl e t o obt ai n t he i nf ormat i on t hey need when t hey need i t ,
t hey must choose bet ween del ayi ng deci si ons or maki ng deci si ons on i ncompl et e and out dat ed
snapshot s of i nf ormat i on.
The probl em f or t he Vi ce Presi dent i s t hat merel y gat heri ng demographi c and operat i onal dat a i n
a warehouse does not ensure ef f ect i ve i nt el l i gence. To be usef ul , she must make t hi s dat a
avai l abl e i n real - t i me f or BI, at t he poi nt of a deci si on. It must be accessi bl e t o t he company’s
hundreds of busi ness users around t he worl d– exact l y when and where t hey need i t .
The company’s BI i nf rast ruct ure i s great l y chal l enged by t he demands of st ori ng and anal yzi ng
i nf ormat i on. It i s a pat chwork of t echnol ogy t hat has accumul at ed f or a vari et y of reasons over a
l ong peri od: many f l avors of general - purpose DBMS sof t ware, a part i al l y i mpl ement ed mi ddl eware
st rat egy or t wo, a col l ect i on of di f f erent mi d- t i er SMP servers, assort ed di sk arrays and myri ad
end- user appl i cat i ons rel yi ng on di f f erent dat abase and communi cat i on st andards. And t hi s
pat chwork of t echnol ogy i s hel d t oget her by expensi ve dat abase and syst em admi ni st rat ors who
are st ruggl i ng t o keep up wi t h user demands.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 3
As more i nf ormat i on i s added t o exi st i ng syst ems, t hey are sl owi ng dramat i cal l y and becomi ng
unrel i abl e. Onl i ne i ni t i at i ves, such as capt uri ng and anal yzi ng cl i ckst ream dat a, t hreat en t o
overwhel m t he i nf rast ruct ure ent i rel y. As BI demands and cust omer dat a grow, t he Vi ce Presi dent
of Cust omer Knowl edge Management and t he CIO are f aced wi t h hi ri ng expensi ve DBAs and
syst ems admi ni st rat ors and maki ng mul t i - mi l l i on- dol l ar i nvest ment s i n hardware, sof t ware,
net worki ng and st orage, i n a l osi ng bat t l e t o mai nt ai n current l evel s of perf ormance.
Thi s si t uat i on i s a common one f or Gl obal 1000 compani es across many i ndust ri es. Over t he past
decade, hi gh- prof i l e management i ni t i at i ves have st rai ned t he current i nf rast ruct ure wi t h t hei r
need f or near- i nst ant access t o dat a across t he ent erpri se:
• Cust omer Rel at i onshi p Management (CRM) i nt egrat es al l areas of busi ness t hat af f ect a
cust omer and requi res det ai l ed, comprehensi ve i nf ormat i on about cust omers across al l t ouch
poi nt s. Thi s i nf ormat i on i s requi red f or coordi nat ed market i ng, sal es and servi ce act i vi t i es.
Ti mel y dat a i s cruci al i f a company i s t o communi cat e consi st ent l y wi t h i t s cust omers.
• Operat i ons management al l ows compani es t o i dent i f y t rends and pat t erns and st reaml i ne
t hei r net work and f i nanci al operat i ons.
• Suppl y Chai n Management (SCM) enabl es a group of compani es t o work t oget her t o opt i mi ze
product assembl y, del i very and i nvent ory. In t he past , manuf act urers cont rol l ed t he pace at
whi ch t hey bui l t and di st ri but ed t hei r product s. Now cust omer orders dri ve t he process and
manuf act urers keep pace wi t h cust omers’ needs f or product f eat ures, order f ul f i l l ment and
f ast del i very.
• Ent erpri se Resource Pl anni ng (ERP) coordi nat es i nf ormat i on f rom depart ment s across t he
ent erpri se, i ncl udi ng manuf act uri ng, order ent ry, account s recei vabl e and payabl e,
purchasi ng, warehousi ng, t ransport at i on and human resources. Compani es rel y on t hei r ERP
syst ems t o hel p t hem anal yze f i nanci al model s, cal cul at e i nvent ory val uat i ons, perf orm
f orecast i ng and t ri gger repl eni shment cycl es.
• Part ner Rel at i onshi p Management (PRM) al l ows compani es t o communi cat e and i nt eract wi t h
t hei r resel l ers and part ners. As wi t h CRM, t i mel y and personal i zed i nf ormat i on l eads t o more
ef f ect i ve and prof i t abl e part nershi ps.
Al l t hese i ni t i at i ves share a need f or more t i mel y, det ai l ed and current dat a f rom a broad range of
sources. They have al l resul t ed i n l ong, cost l y i mpl ement at i ons of sophi st i cat ed t ool s and
appl i cat i ons t o al l ow easi er access and cont rol . Most of t hem requi re bot h ad hoc queri es and
compl ex anal yses, such as dat a mi ni ng, whi ch enabl e compani es t o uncover pat t erns and
rel at i onshi p by di ggi ng t hrough massi ve amount s of dat a.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 4
Wi t h al l t hese st resses on t he current BI i nf rast ruct ure, i t i s not surpri si ng t hat many busi ness and
t echni cal pai n poi nt s have emerged:
• Sl ow access t o i nf ormat i on i nhi bi t s spont aneous queri es, whi ch resul t s i n l ost opport uni t i es
• Compl ex queri es and report s requi re hours or even days t o process
• To speed processi ng, dat a must be sampl ed and summari zed, l i mi t i ng dept h of anal ysi s
• Gat heri ng usef ul i nf ormat i on requi res t rai ni ng. Yet t he peopl e who need t he dat a are t ypi cal l y
not t he same peopl e who can run t he queri es. Thi s creat es bot t l enecks i n t he organi zat i on, and
resul t s i n l ost opport uni t i es.
• Produci ng predi ct i ve model s i s di f f i cul t and of t en requi res a st af f of st at i st i ci ans
• Scal abi l i t y suf f ers as dat abases grow beyond hundreds of gi gabyt es. Cost s i ncrease
unpredi ct abl y and di scont i nuousl y, yet perf ormance remai ns poor.
• Dat a l oadi ng i s sl ow and pai nf ul and requi res a medl ey of ETL t ool s; queri es run on out dat ed
i nf ormat i on can yi el d mi sl eadi ng or i naccurat e resul t s
• Mergi ng or st andardi zi ng dat a i s t i me- consumi ng and di f f i cul t because dat a i s hel d i n a
vari et y of f ormat s and on l egacy syst ems
• Parsi ng of dat a mart s f rom l arge dat a warehouses t o ret ri eve dat a i s di f f i cul t t o achi eve i n an
accept abl e t i me f rame
• Teams of dat abase and syst ems admi ni st rat ors are requi red t o t une syst ems and queri es t o
achi eve accept abl e perf ormance
Mai nf r ame
Dat a
ERP
Appl i c at i ons
CRM
Appl i c at i ons
OLTP
Oper at i onal
Dat a St or age
M
e
t
a
D
a
t
a
M
a
n
a
g
e
m
e
n
t
O
L
A
P
S
e
r
v
e
r
R
e
p
o
r
t
i
n
g
a
n
d
Q
u
e
r
y
F
u
n
c
t
i
o
n
s
D
a
t
a
E
x
t
r
a
c
t
i
o
n
,
T
r
a
n
s
f
o
r
m
a
t
i
o
n
,
a
n
d
L
o
a
d
i
n
g
Dat a
War ehouse
Cust omer
Pr of i t abi l i t y
Fi nanc i al
Pl anni ng
Busi ness
Devel opment
Mar k et i ng Ful f i l l ment I nvent or y
Dat a Mar t s
Mul t i pl e sour c es c ont r i but e t o
i nf r ast r uc t ur e st r esses.
Busi nesses must c onst ant l y
j uggl e new dat a wi t h ar c hi ved
dat a, new sour c es wi t h exi st i ng.
Conf i dent i al
Al t hough BI i s wi del y recogni zed as i mport ant , t oo many i ni t i at i ves have ended as cost l y f ai l ures.
The expl anat i ons somet i mes ment i on ‘ i l l def i ned goal s’ , or ‘ i nsuf f i ci ent commi t ment f rom
management ’ . But of t en, ‘ i nsuf f i ci ent commi t ment ’ means t hat t he i nf rast ruct ure proved
unexpect edl y expensi ve. And f requent l y, ‘ i l l - def i ned goal s’ are goal s t hat were promi sed by a sal es
person, but t hat coul dn’ t be achi eved on t he act ual pl at f orm del i vered. The f undament al probl em
i s of t en t hat t he underl yi ng dat a warehouse i s t oo hard t o set up and admi ni st er, t oo hard t o use,
has perf ormance t hat i s t oo sl ow and i s unabl e t o keep pace wi t h growt h.
The Hi st or y of t he BI I nf r ast r uc t ur e
How di d t hi s exi st i ng BI i nf rast ruct ure evol ve, and why has i t been unabl e t o keep pace? Decades
ago, t he ori gi nal BI i nf rast ruct ure f ocused on dat a warehouses t o archi ve dat a f rom operat i onal
syst ems. The purpose of t he ori gi nal archi ve was t o st ore l arge amount s of dat a f or some
subsequent use, not t o anal yze dat a qui ckl y. Anal ysi s was consi dered l at er.
The ori gi nal i nf rast ruct ure di d not ant i ci pat e t he ki nds of needs t hat woul d ari se 10 t o 15 years
af t er i t s bi rt h. Generi c hardware, i n t he f orm of st orage medi a and f i l e servers, was added as
needed wi t hout seri ous t hought of f ut ure requi rement s and demands. Sof t ware f or servi ng t he
dat a was creat ed based on i mmedi at e needs. The resul t i ng mi x of hardware and sof t ware
creat ed a makeshi f t sol ut i on t hat users cont i nual l y ‘ t weaked’ t o accompl i sh more.
Bef ore 1990, much dat abase dat a was kept on l egacy syst ems, many of whi ch
were mai nf rames runni ng such programs as DB2. Programmers pai nst aki ngl y
wrot e programs t o ext ract i nf ormat i on f rom t he dat a st ored on t hese syst ems.
The i nf rast ruct ure began t o evol ve spont aneousl y, wi t h l i t t l e hi gh- l evel pl anni ng or
consi derat i on f or f ut ure needs.
Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d
The i nf r ast r uc t ur e has been
evol vi ng spont aneousl y, wi t h
l i t t l e hi gh- l evel pl anni ng or
c onsi der at i on f or f ut ur e needs.
Client/Server Era
DBMS
File Servers:
Undifferentiated
Blocks of Binary Data
General Purpose Boxes:
Scientific/Media/Data/Email/Web...
DBMS:
I BM
I nformix
Microsoft
Sybase
Teradata
Oracle
Storage:
EMC
Hitachi
Veritas
Servers:
HP
I BM
Sun
Teradata
End User BI
and Vertical
Applications
The exi st i ng i nf r ast r uc t ur e t r i es t o
mak e do wi t h a pat c hwor k of
har dwar e, sof t war e and st or age,
some of whi c h was never meant t o
wor k t oget her seaml essl y.
5
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 7
In t he 1990s, personal comput ers gave advanced users access t o t he l egacy syst ems and enabl ed
t hem t o perf orm anal yses. Busi ness users f ound t hat t he dat a t hey coul d now access was
f ragment ed, and came i n many f ormat s across dat abases, f i l es and spreadsheet s.
Dat a use was di vi ded i nt o t wo pri mary act i vi t i es: sl i ci ng and anal yzi ng dat a at one l evel , and
dri l l i ng down i nt o dat a. Response t i me was a seri ous i ssue f or a dat a warehouse of any
si gni f i cant si ze. Throughout t he 1990s, comput er hardware and sof t ware dropped i n pri ce and
i ncreased si gni f i cant l y i n perf ormance. Anal ysi s of l arge amount s of dat a became more f easi bl e
and desi rabl e. Cl i ent / server comput i ng t ook hol d, al ong wi t h report i ng and anal ysi s t ool s,
al l owi ng PC users t o access dat a warehouses more easi l y and f rom di f f erent geographi cal
l ocat i ons. Response t i me and st abi l i t y were st i l l i ssues. By 2000, t he BI i nf rast ruct ure t hat had
evol ved— a poorl y bui l t f oundat i on— was now t asked wi t h support i ng a compl ex, growi ng set of
t ool s, appl i cat i ons, users, processes and demands.
Recent l y, a number of f orces have come i nt o pl ay whi ch l ead t owards a new BI pl at f orm, one t hat
addresses t he probl ems of t he ol d i nf rast ruct ure. The f i rst i s t he wi despread use of open, i ndust ry
st andard i nt erf aces across pl at f orms, i ncl udi ng ODBC, JDBC and ot hers. The second i s t he
rapi dl y decl i ni ng cost and i ncreasi ng perf ormance of di sk, processi ng, net work and st orage
component s— f or exampl e, t he wi del y known Moore’s Law t hat st at es t hat dat a densi t y per square
i nch on i nt egrat ed ci rcui t s i s doubl i ng approxi mat el y every 18 mont hs. The t hi rd i s t he
exponent i al growt h of dat a, est i mat ed by Greg’s Law t o doubl e on average every ni ne mont hs.
Fi nal l y, t here i s t he emergence and adopt i on of speci al - purpose appl i ances, i ncl udi ng net work
appl i ances, medi a servers and ot hers. Toget her, t hese f orces open a door t o a pl at f orm t hat of f ers
revol ut i onary pri ce/ perf ormance f or t era- scal e BI.
Towar ds A New BI Pl at f or m
Today, a new BI pl at f orm i s needed t o rat i onal i ze and revi t al i ze t he exi st i ng i nf rast ruct ure. Thi s new
pl at f orm provi des a f oundat i on f or t he exponent i al dat a growt h t hat i s common i n t oday’s
busi ness envi ronment . It combi nes massi vel y paral l el hardware, sof t ware and st orage and i s
di rect l y f ocused on provi di ng opt i mal response t i mes and scal abi l i t y at t he t erabyt e l evel . The new
BI pl at f orm el i mi nat es yest erday’s pat chwork of hardware, sof t ware and st orage, and enabl es
opt i mi zed access t o i nf ormat i on. As busi ness and t echni cal demands cont i nue t o grow
and change i n t he new cent ury, t he new BI pl at f orm i s desi gned t o scal e wi t h dat a
si ze, scope and perf ormance needs. And i t does al l t hi s at an af f ordabl e and
predi ct abl e pri ce.
The new pl at f orm of f ers an open archi t ect ure, massi vel y paral l el processi ng and
scal abi l i t y and al l ows growt h wi t hout compl ex upgrades or pai nf ul i nt egrat i ons of
new syst ems.
The new BI pl at f or m el i mi nat es
yest er day’ s pat c hwor k of har dwar e,
sof t war e and st or age, and enabl es
opt i mi zed ac c ess t o i nf or mat i on.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 7
As dat a grows t o hundreds of t erabyt es, a company can seaml essl y add t he hardware
necessary f or addi t i onal i nf ormat i on. Dat a i s mi rrored f or prot ect i on agai nst l oss, and t here i s hi gh
avai l abi l i t y t hrough redundancy. Rat her t han over- usi ng one huge dat abase, t he company can
have a speci f i cal l y engi neered, t era- scal e dat a warehouse. Al l dat a can be l i nked and managed
f rom one l ocat i on. Thi s saves admi ni st rat i ve resources and reduces cost s t hrough bet t er
deci si on- maki ng and t he need f or f ewer syst ems and dat abase admi ni st rat ors.
For bot h busi ness and t echni cal users, t here are many benef i t s of t he new BI pl at f orm. Si nce
l egacy syst ems and exi st i ng BI t ool s i nt egrat e easi l y wi t h t he new pl at f orm, busi ness users can
perf orm al l t hei r current anal yses wi t h f ami l i ar i nt erf aces— but i n a f ract i on of t he t i me, and wi t h
t he opport uni t y t o add new BI act i vi t i es t hat previ ousl y woul d have been i mpossi bl e. The new
pl at f orm al l ows a qui cker response t o cust omer, part ner and empl oyee needs. It ensures
perf ormance, rel i abi l i t y and scal abi l i t y, and reduces t he st rai n and demands on corporat e IT.
There are as many benef i t s of t he new BI pl at f orm as t here are di verse busi nesses. Wi t h real - t i me
access t o vast st ores of BI:
• Tel ecommuni cat i ons carri ers can anal yze t erabyt es of cal l det ai l records (CDRs) i n real - t i me
t o predi ct prof i t abl e cust omers at ri sk t o churn and t ake prevent at i ve act i on, savi ng bi l l i ons
of dol l ars
• Ret ai l ers can run comprehensi ve CRM anal yses t o det ermi ne whi ch on- t he- spot deal s
wi l l boost revenues
• Fi nanci al servi ces i nst i t ut i ons can run f ast , comprehensi ve credi t ri sk anal yses f or l oan
approval s, savi ng mi l l i ons i n l oan def aul t s
• E- busi nesses can det ermi ne t he short est cl i cks- t o- purchase pat h t hrough
cl i ckst ream anal yses
• Out sourced anal yt i c servi ce provi ders can real i ze a compet i t i ve advant age by of f eri ng f ast er
response t i mes i n t hei r SLAs
• Credi t card compani es can det ect and handl e f raud as soon as i t happens by i mmedi at el y
pi cki ng up i nconsi st enci es i n buyi ng habi t s or geography. They can uncover new f raud
pat t erns and qui ckl y appl y rul es t o el i mi nat e t hem.
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 8
The Net ezza Syst em
The Net ezza Perf ormance Server™ 8000 seri es i s t he real i zat i on of t he new BI pl at f orm. The
Net ezza Perf ormance Server (NPS™ ) syst em i s an ent erpri se- cl ass dat a warehouse appl i ance t hat
del i vers breakt hrough perf ormance and ease- of - use at a f ract i on of t he cost of t radi t onal dat a
warehouses. The NPS appl i ance of f ers:
• 10 t o 50 t i mes t he per f or manc e of exi st i ng dat a war ehouse syst ems at hal f t he pr i c e—
f i nal l y an af f ordabl e sol ut i on t o busi ness users’ perf ormance needs
• Ti ght i nt egrat i on of st orage, server and dat abase component s, and i t s massi vel y paral l el
archi t ect ure provi des rapi d, opt i mi zed execut i on of i nt eract i ve queri es, as wel l as compl ex
report s and anal yses
• Ext reme ease- of - use and rel i abi l i t y— a part - t i me DBA i s al l t hat ’s requi red
t o manage t he syst em.
• Real - t i me dat a l oadi ng, sel f - t uni ng dat a st orage and hi gh t hroughput on l arge,
compl ex workl oads
• St rai ght f orward i nt egrat i on, t hrough i t s open APIs, wi t h al l maj or el ement s of t he BI
envi ronment : BI appl i cat i ons and t ool s, ETL and EAI t ool s, l egacy syst ems and dat a sources.
The Net ezza Per f or manc e Ser ver
appl i anc e i s t he c or e of f ast ,
c ompr ehensi ve BI anal yses,
r egar dl ess of dat a si ze.
Other Data Sources
Operational
Business
Systems
Programmers
Designers
Administrators
Data Mining
Business Analysis & Marketing
I mplementation
Decisions
Business Partners
Ecommerce
Vendors
Customers
Netezza
Performance
Server
System
Conf i dent i al Bu s i n e s s I n t e l l i g e n c e i n a Re a l - Ti me Wo r l d 9
The NPS syst em opt i mi zes BI queri es f or maxi mum perf ormance. The archi t ect ure i s desi gned f or
speed and i s l i mi t ed onl y by di sk I/ O. The NPS syst em i s bui l t f or qui ck depl oyment , si mpl e
admi ni st rat i on and nearl y i nf i ni t e scal i ng— i nt o t he hundreds of t erabyt es. The t ot al cost
of ownershi p (TCO) of Net ezza’s perf ormance i s a f ract i on of t he cost of sol ut i ons
current l y on t he market .
The dramat i c perf ormance of f ered by t he NPS syst em enabl es real - t i me
f orecast s and anal ysi s. ‘ What i f ’ scenari os can be perf ormed by t he peopl e who
use t he dat a, rat her t han havi ng t o wai t i n a queue f or IT support . Queri es t hat
used t o t ake hours are reduced t o a f ew seconds, so t hat compani es can respond
proact i vel y t o opport uni t i es i nst ead of react i ng t o probl ems. The bot t om l i ne:
easi er and f ast er access t o i nf ormat i on hel ps busi ness users make bet t er deci si ons.
Conc l usi on
Ti mel y access t o BI has never been more i mport ant . Sophi st i cat ed use of BI can hel p organi zat i ons
reduce cost s, pl an i nvent ory, i dent i f y sal es and market i ng opport uni t i es, opt i mi ze pri ci ng and,
ul t i mat el y, i ncrease revenues. For many compani es, aggressi ve and real - t i me BI act i vi t i es can
make t he di f f erence bet ween wi nni ng and l osi ng t o t he compet i t i on.
Wi t h so much at st ake, ent erpri ses can no l onger af f ord t o i mprovi se sol ut i ons. A pat chwork
syst em of hi gh- end hardware, general - purpose dat abase sof t ware and st orage may cost a l ot of
money, but i t cannot consi st ent l y del i ver real - t i me perf ormance f or rapi dl y growi ng, t era- scal e
dat a st ores. In order t o achi eve t he promi se of BI, a new pl at f orm i s requi red.
The NPS syst em i s desi gned f or a si ngl e purpose: t o del i ver t era- scal e BI 10 t o 50 t i mes f ast er t han
exi st i ng dat a warehouse sol ut i ons, at hal f t he cost . Wi t h t he NPS syst em’s unbeat abl e
pri ce/ perf ormance, l ow TCO and ease- of - use, compani es across a broad range of i ndust ri es are
f i nal l y abl e t o l everage t hei r Busi ness Int el l i gence and real i ze a compet i t i ve advant age.
WP.BIRTW.1.04
Net ezza Cor por at i on 200 Crossi ng Boul evard, Frami ngham, Massachuset t s 01702- 4480
t el 508.665.6800 f ax 508.665.6811 www.net ezza.c om
© Net ezza 2004. Al l ri ght s reserved. Net ezza, t he Net ezza l ogo, NPS, Perf ormance Server, Net ezza Perf ormance Server, Int el l i gent Query St reami ng,
Asymmet ri c Massi vel y Paral l el Processi ng and AMPP are t rademarks or regi st ered t rademarks of Net ezza Corporat i on i n t he U.S. and/ or ot her count ri es.
Al l ot her company, brand and product names cont ai ned herei n may be t rademarks or regi st ered t rademarks of t hei r respect i ve hol ders.
The Net ezza Per f or mance Ser ver
syst em i s t he f i r st ent er pr i se- c l ass
dat a war ehouse appl i anc e opt i mi zed
f or BI on t ens, even hundr eds,
of t er abyt es of dat a i n t oday’s
r eal - t i me busi ness wor l d.
doc_570265564.pdf