VLDB Challenges VLDB Challenges in in Very Large Very Large - - PowerPoint PPT Presentation
VLDB Challenges VLDB Challenges in in Very Large Very Large - - PowerPoint PPT Presentation
VLDB Challenges VLDB Challenges in in Very Large Very Large Enterprises Enterprises Panelists Chair Chair Dr. Michael L. Brodie, Chief Scientist, Verizon Problem Owners Problem Owners Dr. Hans-Peter Steiert, Research &
Chair Chair
- Dr. Michael L. Brodie, Chief Scientist, Verizon
Problem Owners Problem Owners
- Dr. Hans-Peter Steiert, Research & Technology,
DaimlerChrysler AG
Solution Owners Solution Owners
- Adam Bosworth, VP, Engineering, BEA Systems Inc
- James Hamilton, Architect, Microsoft SQL Server, Microsoft
- Pat Selinger, IBM Fellow and VP, Data Management
Architecture and Technology, IBM
Panelists
Very Large Enterprises
Petabyte
248 $ 6 7 1 1 2 6 Verizon 373 $ 1 3 6 N/ A 7
Daim ler Chrysler
Data
Em ployees ( 1 ,0 0 0 s)
Revenues ( $ B US)
Fortune 5 0 0 Global 5 0 0 Very Large Enterprise
Problem Drivers Problem Drivers
- Data Growth
- Data Life Cycle
OLTP W orkload Grow th
1,000 2,000 3,000 Jan-01 Dec-02 Dec-04
OLTP/sec Triples 2001-2004
Projected average workload
0% 40% 80% 120% Two Years 1/01 - 12/02 Four Years 1/01 - 12/04OLTP Doubles by 12/04
Projected Workload Growth Rate (average)DSS W orkload Grow th
25 50 75 100 125 150 On 1/2001 On 12/2002 On 12/2004
Inflight Queries Double by 04
Projected workload (average) (concurrent inflight queries)
0% 20% 40% 60%
2 Yr Growth 4 Yr growth
DSS Workload Triples by 04
Projected workload growth rate (average)
Database Grow th
1 2 3 4 5 6
Respondents' Projected Database Size (average) (TB)
On Jan 2001 On 12 2002 On 12 2004
Size: OLTP Doubles; DSS Triples by 04
OLTP DSS OLTP N = 43 DSS N = 67
0% 50% 100% 150%
Jan 01 - Dec 02 Jan 01 - Dec 04
Growth: DSS & OLTP Double by 04
OLTP DSS
Respondents' Projected Database Growth Rate (average)
VLE Storage Grow th
500 1000 1500 2000 2500 3000 3500 Terabytes 2001 2002 2003 2004 2005 2006 Calendar Year
VLE X Storage Triples 02-06
IP SAN/NAS OS Vertical OS Shared FC SAN System390
Data Life Cycle
Life Cycle Actions Life Cycle Actions
- Create
- Store
- Replicate
- Protect
- Update
- Archive
- Exchange, exchange, exchange, …
Factors Factors
- History: 40+ years of Mergers & Acquisitions
- Growth: Automation & Partnering
- Protection: security, confidentiality, …
- The Grand Challenge: semantics of data
Data Droppings Problem
Grand Challenge: $ 1 Trillion/ year
Integration Cost Estimates Integration Cost Estimates
- 24% of IT budgets: $180 B / year US (InfoWorld, January 2002)
- 13% of IT spend: $752 B / year US (Giga estimate; May 2002)
- 25-40% of all IT projects (various)
- 6% of US IT spending: $610B / year US (IDC, May 2002)
- 7% of IT spending: $1.3T / year worldwide (IDC, May 2002)
- 28+% of worldwide consulting: $ 160 B/year (Gartner, March 2002)
- 43% of e-business worldwide consulting: $53 B / year (Gartner)
- 1.75% to annual IT budget on EAI and B2Bi (Forrester, Dec 2001)
- 10-30% of IT budgets (David Sink, IBM, InformationWeek, May 27,
2002)
Data Quality Cost Estimates Data Quality Cost Estimates
- $600 B / year US (Data Warehouse Institute, 2002)
VLE VLDB Challenges
Data Management Data Management
- Global Data Managem ent
– Significant improvement in dealing automatically with semantics
- Database Engineering
- Automated DBA
- Com prehensive Data Managem ent Architecture
- Data architecture: Web Services, mid-tier, distributed data
Storage Management Storage Management
- Data Protection
– Integrated products: DBMS, replication, …
- Data Utilization
– Automated DBA, Storage Virtualization, Hierarchical Storage Management for distributed system