Database Systems Laboratory -- Projects
Workload Management for Database Management Systems
DBMSs are responsible for running multiple workload types with varying resource usage patterns and associated Service Level Objectives (SLOs). The role of the DBMS is to process all incoming work while meeting the SLOs of each workload. In order to do so efficiently, various workload management techniques are employed. We have explored several types of workload management in the past (query throttling, resource allocation control mechanisms, scheduling) and we are currently developing an architecture to determine the appropriate strategy to use given current system conditions. This work is being carried out with our industrial partner, IBM Canada.
Management of Elastic Services in the Cloud
Cloud computing, with its support for elastic resources that are available on an on-demand, pay-as-you-go basis, is an attractive platform for hosting Web-based services that have variable demand, yet consistent performance requirements. Effective service management is mandatory in order for services running in the cloud, which we call elastic services, to be cost-effective. We are developing a management framework to facilitate elasticity of resource consumption by services in the cloud. This framework is based on a previous framework designed for services management with the necessary concepts and properties to support elastic services. This work is joint work with the Universidad Complutense de Madrid in Madrid, Spain.
Delivering Ultra-large-scale Services
Ultra-large Scale (ULS) services (eg. communication platforms, social networking sites, e-banking etc) must cope with problems of scale that dwarf the traditional issues facing modern software systems. ULS services process the critical (e.g., financial and healthcare) data of millions of people, creating the potential for security and privacy breaches on an unprecedented scale. They are deployed on massive and costly computing infrastructures, and they must handle millions of users performing billions of transactions on a daily basis without ever breaking. Outages of ULS services like MobileMe, Gmail, Blackberry, Interac or PayPal can bring economies to a halt and are often headline news.
Current industrial approaches to cope with the characteristics of ULS services are usually ad hoc, costly last-resort efforts. The proposed research program will develop principled approaches to provide and evolve high-quality services in a cost-effective manner while handling their continuously growing user base. The research program will achieve its goal by:
The research team consists of leading academic experts in software engineering, networking, security, autonomic computing, monitoring and databases from Queen’s University, the University of Waterloo and the University of Western Ontario. We work closely with the providers of some of the world’s largest and most complex ULS services (Bell Canada’s wireless network and RIM’s Blackberry platform) to tackle important and practical problems. Our results will provide the knowledge for companies across Ontario (Amika Mobile, Christie Digital, IBM Canada) to make their products ULS-ready, to ensure that their customers can play a pivotal role in ULS service deployments, and to maintain and grow Ontario’s competitive advantage in the ULS domain.
Read More
Current industrial approaches to cope with the characteristics of ULS services are usually ad hoc, costly last-resort efforts. The proposed research program will develop principled approaches to provide and evolve high-quality services in a cost-effective manner while handling their continuously growing user base. The research program will achieve its goal by:
- Developing autonomic techniques to reduce the cost of operating the infrastructures of ULS services. Infrastructures that adapt to the demand characteristics of the services that run on them would ensure that services are provisioned in the most cost-effective manner.
- Creating automated approaches to verify the scalability of ULS services. Verification approaches would ensure that ULS services can function reliably in a secure and cost-effective fashion.
- Investigating novel mechanisms to deliver ULS services in a distributed and flexible fashion. Flexible delivery mechanisms will help in sharing the cost of service delivery with users instead of centralizing the processing and costs in a limited number of data centers.
- Exploring novel approaches to quantify service quality in a ULS setting. Such approaches will reduce the costs of delivering high quality services by focusing quality improvement efforts on areas where quality matters the most to the users of ULS services.
The research team consists of leading academic experts in software engineering, networking, security, autonomic computing, monitoring and databases from Queen’s University, the University of Waterloo and the University of Western Ontario. We work closely with the providers of some of the world’s largest and most complex ULS services (Bell Canada’s wireless network and RIM’s Blackberry platform) to tackle important and practical problems. Our results will provide the knowledge for companies across Ontario (Amika Mobile, Christie Digital, IBM Canada) to make their products ULS-ready, to ensure that their customers can play a pivotal role in ULS service deployments, and to maintain and grow Ontario’s competitive advantage in the ULS domain.
Read More
CASTLE - A Training Oriented Adaptive Decision Support System for DBAs
Decisions are made based on knowledge. We are working on a framework that dynamically extracts knowledge from various correlated data sources containing systems related data and from the problem solving strategies of the expert Mainframe DB2 Database Administrators (DBA). The framework then uses the knowledge to train the new generation of DBAs by guiding them through the various stages of solving DB2 problems on the Mainframe system. The research combines text and data mining techniques for knowledge extraction, a rule-based system for knowledge representation and problem categorization, and a case-based system for decision support. The framework provides an interactive interface to accept user preferences, which is used to adapt the rule and case-bases. Rules are extracted from various log data sources and monitoring data in the data warehouses. The rule-based system guides the user through the initial trivial steps of problem investigation and categorization, and helps collect more information about the problem. Finally the case-based system is searched for a suitable solution. This work is being done in collaboration with our industry partners, CA Technologies.
Self-Managing Cloud Data Services
José Luis Vázquez-PolettiCloud data services, which are the cloud computing versions of
traditional database management systems (DBMSs), are intended to
offer data persistence, management and retrieval as services from
the cloud. Current cloud data services, however, offer very
restricted functionality compared with traditional DBMSs largely due
to the complexity inherent in managing these services. The
complexity stems from basic features of cloud data services such as
multi-tenancy, high variability in the workload, distributed data
and elastic resources.
We are addressing the problem of the poor manageability of cloud data services by advancing the creation of self-managing cloud data services. We will be developing novel models and methods to support adaptable data partitioning, migration and replication, and will be providing a framework for flexible and adaptable data consistency.
This project is in collaboration with José Luis Vázquez-Poletti at the Complutense University of Madrid (Spain).
We are addressing the problem of the poor manageability of cloud data services by advancing the creation of self-managing cloud data services. We will be developing novel models and methods to support adaptable data partitioning, migration and replication, and will be providing a framework for flexible and adaptable data consistency.
This project is in collaboration with José Luis Vázquez-Poletti at the Complutense University of Madrid (Spain).
