Scribd
Upload
46 views10 pages

Physical Design and Governance of Data Warehouses: Lesson 1: Storage Architectures

This document compares physical data warehouse architectures and storage structures. It discusses multidimensional online analytical processing (MOLAP), relational online analytical processing (ROLAP), and hybrid online analytical processing (HOLAP). Bitmap indexes and columnar storage are explained as optimized structures for star schema queries and business intelligence workloads. Columnar storage compresses and sorts columns, improving performance of analytics on large fact tables over traditional row storage.

Uploaded by

anchal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
46 views10 pages

Physical Design and Governance of Data Warehouses: Lesson 1: Storage Architectures

This document compares physical data warehouse architectures and storage structures. It discusses multidimensional online analytical processing (MOLAP), relational online analytical processing (ROLAP), and hybrid online analytical processing (HOLAP). Bitmap indexes and columnar storage are explained as optimized structures for star schema queries and business intelligence workloads. Columnar storage compresses and sorts columns, improving performance of analytics on large fact tables over traditional row storage.

Uploaded by

anchal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 10

Information Systems Program

Module 5
Physical Design and Governance
of Data Warehouses

Lesson 1: Storage Architectures


Lesson Objectives
• Compare and contrast physical architectures for
implementing data warehouses
• Explain components of bitmap join indexes for
star join queries
• Explain advantages of columnstore structures for
business intelligence queries

Information Systems Program


Storage Architectures
Query Performance
Low Medium High

ROLAP HOLAP MOLAP

Dim 1 Dim 1
Cubes
MVs Cubes
Fact 1 Fact 1 MVs

Capacity
High Medium Low
3

Information Systems Program


Bitmap Index
 Useful for stable columns with few values
 Bitmap:
– String of bits: 0 (no match) or 1 (match)
– One bit for each row: 110010010001
 Bitmap index record
– Column value: “USA”
– Bitmap: 110010010001
– Convert a bit position into a row identifier

Information Systems Program


Bitmap Index Example
SSCustomer
RowId CustNo … CustCountry
1 098-55-1234 USA
2 123-45-6789 USA
3 456-89-1243 Canada
4 111-09-0245 Mexico Bitmap Index on CustCountry
5 931-99-2034 USA CustCountry Bitmap
6 998-00-1245 Mexico USA 110010010001
7 287-44-3341 Canada Canada 001000100100
8 230-21-9432 USA Mexico 000101001010
9 321-44-5588 Mexico
10 443-22-3356 Canada
11 559-87-3211 Mexico
12 220-44-5688 USA

Information Systems Program


SSSales Bitmap Join Index Example
RowId SalesNo … CustNo
1 1111   098-55-1234
2 1234   123-45-6789
3 1345   456-89-1243
4 1599   111-09-0245
5 1807   931-99-2034
6 1944   998-00-1245
Bitmap Join Index on CustCountry
7 2100   287-44-3341 CustCountry Bitmap
8 2200   230-21-9432 USA 110010010001110010010001
9 2301   321-44-5588 Canada 001000100100001000100100
10 2487   443-22-3356 Mexico 000101001010000101001010
11 2500   559-87-3211
12 2600   220-44-5688
13 2703   098-55-1234
14 2801   123-45-6789
15 2944   456-89-1243
16 3100   111-09-0245
17 3200   931-99-2034
18 3258   998-00-1245
19 3302   287-44-3341
20 3901   230-21-9432
21 4001   321-44-5588
22 4205   443-22-3356
23 4301   559-87-3211 6
24 4455   220-44-5688

Information Systems Program


Row Storage versus Column Storage
Row Storage
• Entire rows in physical
records
• Full table scans for business
intelligence queries
• Foundation storage approach
for most enterprise DBMSs

Column Storage
• Specialized storage structure
• Eliminates table scan for
typical business intelligence
queries
• Optional storage approach in
enterprise DBMSs

Information Systems Program


Columnstore File Structure
Row Group i
Col1 File Col2 File ... Colk File
Col1 value 1 Col2 value 1 Colk value 1
Col1 value 2 Col2 value 2 Colk value 2
… … …

Col1 value m Col2 value n Colk value p

Information Systems Program


Columnstore Index
• Store columns in physical records
• Compression for duplicate values
• Optional sorting

Sales fact row group Columnstore indexes


SalesNo SalesUnits SalesAmt SalNo SalesUnits SalesAmt
1111 10 1200 1111 10× 2 850
1122 20× 2 900
1122 20 900
1133 850
1133 50 850 1144 50× 2
1150
1144 10 1400 1155 1400 × 2
1155 20 1150 1166
1166 50 1400

Information Systems Program


Summary
• Basic architectures: MOLAP, ROLAP, HOLAP
• Bitmap indexes for stable columns with few
values
• Columnar storage becoming common for large
fact tables

10

Information Systems Program

You might also like