Transcript IBM Presentations: Grid Basic template

Almaden Research Center
QoS-based Data Access and Placement
for Federated Systems
Wen-Syan Li, Vishal S. Batra,
Vijayshankar Raman, Wei Han,
Inderpal Narang
Grid Computing
Goal: QOS guarantees for database workloads
 E.g., run this query within 30 mins (user’s lunch break)
– Automatically configure dbms, provision resources, etc to meet this
 Very hard problem; demo presents 1st baby step
 Techniques to meet QoS
– Choose among alternate data sources
 At runtime, based on calibrated loads
– Create new replicas
 Based on what-if analysis
– No extra credit for overperforming
 Demo
– QoS aware query routing
– QoS aware data placement
– Availability aware query routing
2
Grid Computing
II Architecture with On Demand Data Placement
Queries
Query Patroller
Explain Table:
Execution plan
Processing cost
Information
Integrator
Simulated Information Grid
Log:
Query statements
Response time
Query
Cost
Calibrator
Query fragment &
response time
Source 1
Source 2
Simulated Meta-Wrapper
Simulated Wrappers
Simulated catalogs
Meta-Wrapper
Wrapper
Simulated
Information
Integrator
Metadata
Repository
Wrapper
Source 3
Data
Placement
Advisor
Virtual tables
Source 1
Source 3
Source 2
Data
Replication
3