Upload
dinesh
View
217
Download
0
Embed Size (px)
Citation preview
8/16/2019 Data WarehouseAima First
1/34
8/16/2019 Data WarehouseAima First
2/34
8/16/2019 Data WarehouseAima First
3/34
8/16/2019 Data WarehouseAima First
4/34
8/16/2019 Data WarehouseAima First
5/34
Subject oriented
8/16/2019 Data WarehouseAima First
6/34
Subject oriented
8/16/2019 Data WarehouseAima First
7/34
8/16/2019 Data WarehouseAima First
8/34
Integrited
8/16/2019 Data WarehouseAima First
9/34
8/16/2019 Data WarehouseAima First
10/34
Time Varient
8/16/2019 Data WarehouseAima First
11/34
Time Varient
8/16/2019 Data WarehouseAima First
12/34
8/16/2019 Data WarehouseAima First
13/34
8/16/2019 Data WarehouseAima First
14/34
8/16/2019 Data WarehouseAima First
15/34
8/16/2019 Data WarehouseAima First
16/34
8/16/2019 Data WarehouseAima First
17/34
8/16/2019 Data WarehouseAima First
18/34
8/16/2019 Data WarehouseAima First
19/34
8/16/2019 Data WarehouseAima First
20/34
8/16/2019 Data WarehouseAima First
21/34
8/16/2019 Data WarehouseAima First
22/34
Architecture of DataWareHouse
8/16/2019 Data WarehouseAima First
23/34
Architecture of a DataWarehouse with a Staging Area
Architecture of a Data
8/16/2019 Data WarehouseAima First
24/34
Architecture of a DataWarehouse with a Staging Area
and Data Marts
8/16/2019 Data WarehouseAima First
25/34
Distributed Data WareHouse
8/16/2019 Data WarehouseAima First
26/34
Incorrect Data in the Datawarehouse.
• The architect needs to know what is to doabout incorrect data in the data warehouse.
• The rst assumtion is that incorrect data
arri!es in the data warehouse on ane"cetion basis.
• If the data is being incorrect#$ entered inthe data warehouse on a who#esa#e basis%
then• It is the dut$ of the architect to nd the
o&ending and make adjustment.
8/16/2019 Data WarehouseAima First
27/34
How to correct
• To correct the o&ending an architect can dothree things.
• '"am#e( suose on ju#$ ) *s +,, is made into oerationa# s$stem on ju#$ - a snashot
taken in data warehouse and on ju#$ )+ itdisco!ered that it was a entr$ of -+, ratherthan +,, on ju#$ ).
• Then
• choice ). go back to ju#$ - and udate -+,insite of +,,. but it can create rob#em if an$reort has been taken between ju#$ - to ju#$ )+.
8/16/2019 Data WarehouseAima First
28/34
How to correct
• choice -.
• 'nter o&setting entr$ i.e make twoentr$ rst debit +,, then credit -+,.
some time it a#so can create rob#em.• hoice /.
• *eset the account to the roer !a#ue.
•
but it wi## not correct the error.• So deending on the situation $ou can
make an$ decision.
8/16/2019 Data WarehouseAima First
29/34
Structuring Data in DataWarehouse
• The sim#est and most common data structurefound in the data warehouse is The sim#estcumu#ati!e structure i.e dai#$ transactions beingreorted from the oerationa# en!ironment.
• '"am#e( jan )% jan- jan/ data• *o##ing summar$ data
• After that the$ are summari0ed into data warehouse records%
• '"am#e( *o##ing summar$ data• Week) data% week- data% month) data month-
data.
8/16/2019 Data WarehouseAima First
30/34
*eorting and the architecteden!ironment
• 1nce the data warehouse has been constructed a##reorting and infromationa# rocessing wi## be donefrom there.
• ). 1erationa# reorting for c#erica# #e!e#• It focus on the #ine item2detai#ed information3.• '"am#e( A cashier has to check who#e da$
transaction in the e!ening for ba#ance check.• -. Data ware house reorting for management #e!e#.• It focus on summar$ information.•
'"am#e( A bank !ice resident has to take decisionhow man$ ATM machine has to #ace in thatarticu#ar cit$ so he does not need one da$transactions but he needs one month or one $earsummar$ of data to take decision.
8/16/2019 Data WarehouseAima First
31/34
4urging Ware house Data
• Data urging is nothing but de#eting$our data from DW.
• Data does not just our into a dataware house. It has its own #ife c$c#ewithin the data warehouse.
• It does not means it is fu##$ remo!edit means it ro##ed u to high #e!e# ofsummar$. Where detai#s is #ost.
8/16/2019 Data WarehouseAima First
32/34
5ranu#arit$
• *efers to the #e!e# of detai#s of the Data
• Dua# #e!e# of 5ranu#arit$(6
• ). 7ow 7e!e# of Detai#2More detai#s3
• -. High 7e!e# of detai#2 #ess detai#s i.eSummar$3
• Most#$ Data in Data warehouse is in High #e!e#
• 8ut it has 7ow 7e!e# of Detai# a#so for atomic9uer$.
8/16/2019 Data WarehouseAima First
33/34
Data 5ranu#arit$
• Data Granularity• A signicant di&erence between an oerationa# s$stem
and a data warehouse is the granu#arit$ of the datastored.
• An oerationa# s$stem t$ica##$ stores data at the
#owest #e!e# of granu#arit$( the ma"imum #e!e# of detai#.• Howe!er% because the data warehouse contains data
reresenting a #ong eriod in time% sim#$ storing a##detai# data from an oerationa# s$stem can resu#t in ano!erworked s$stem that takes too #ong to 9uer$.
• A data warehouse t$ica##$ stores data in di&erent#e!e#s of granu#arit$ or summari0ation% deending onthe data re9uirements of the business. If an enterriseneeds data to assist strategic #anning% then on#$ high#$summari0ed data is re9uired.
8/16/2019 Data WarehouseAima First
34/34
5ranu#arit$
• The #ower the #e!e# of granu#arit$ of data re9uired b$the enterrise% the higher the number of resources2secica##$ data storage3 re9uired to bui#d the datawarehouse. The di&erent #e!e#s of summari0ation inorder of increasing granu#arit$ are(
• urrent oerationa# data• Historica# oerationa# data• Aggregated data• Metadata• urrent and historica# oerationa# data are taken%
unmodied% direct#$ from oerationa# s$stems.Historica# data is oerationa# #e!e# data no #onger9ueried on a regu#ar basis% and is often archi!edonto secondar$ storage.