Collaborative Analytical Processing - Dream or Reality?
Tuesday 11th September, 16:30-18:00, Auditorium
Chair: Bill O'Connell (IBM)
Panelists: Vaishnavi Anjur (Hyperion), Goetz Graefe (Microsoft), and Andy Witkowski (Oracle)
Majority of current Business Intelligence Data (BID) is spread out between incompatible tools like RDBMSs, OLAP engines, and spreadsheets. Bridges between them allow interchange of data and metadata thus providing a small degree of BID sharing, but due to lack of metadata standards, this sharing is limited to specific installations. This creates an unacceptable situation where an analyst cannot get a complete picture of a business. A solution is needed, the Collaborative Analytical Processing (CAP) solution, which imposes a very tight integration between the tools so (meta) data interchange, change management, scalability and data availability is as good as in RDBMSs, and performance of OLAP queries is as good as in the specialized OLAP engines. Any CAP solution must assist business tools through simplification and integration so they can focus on business models, relationship management, fact discovery and presentation than on availability, performance and scalability.
This panel will discuss feasibility of such solution and its place in the spectrum between RDBMS and specialized OLAP engines. Topics will include:
- Should we add new objects to SQL like cubes and dimensions?
- How should standard metadata be represented? Should it be in new SQL objects or standards like CWM?
- How much computation should be done in the tool vs. RDBMs?
- Identify type of queries that OLAP tools will generate against RDBMs
- Identity new types of optimizations needed for these queries
- Identify new execution and access methods to execute these queries
- Identify incremental maintenance issues for CAP where all the data may or may not be co-resident
- How soon will we hit the limits of SQL in terms of expressibility?
Are Web services the Next Revolution in E-Commerce?
Wednesday 12th September, 17:00-18:30, Aula Magna
Chair: Dick Tsur (BEA Systems)
Panelists: Serge Abiteboul (INRIA and Xyleme), Rakesh Agrawal (IBM Almaden Research Center), Umesh Dayal (Hewlett-Packard Laboratories), Johannes Klein (Microsoft Research), and Gerhard Weikum (University of the Saarland)
Electronic commerce in its different manifestations is spreading and rapidly affecting all of us. In most cases we, as individuals, come in touch with it when we purchase goods or services over the Web. This practice is usually referred to as "B2C" or business to consumer interaction. Yet, this is only the tip of the iceberg: the real revolution (to use a good cliché) will come when businesses will electronically conduct their business with other businesses over the Web. This is known as "B2B" or business to business, and is expected to introduce enormous efficiencies into the market place. Today, the trend is shifting from B2C to B2B and a number of leading companies are already engaged in B2B activities: companies such as Ariba and CommerceOne offer to create and operate marketplaces. Other companies such as BEA, IBM, Vitria, Tibco and others offer the infrastructure software and platforms for these marketplaces. In addition, there is an intensive activity in the creation of standards to enable the exchange of information at the B2B level. According to a recent Gartner report, $400 Billion worth of B2B commerce was transacted during the year 2000 and this number is expected to double this year. Web services (or their close relative, E-services) are presently touted to become the next infrastructure technology that at long last, will unleash these efficiencies and in the process, will transform the Web as we know it today, into a distributed application-to-application network with all of the trappings of peer-to-peer architecture.
Panel members who from industry and from academia, have been asked to state their positions on the following issues:
- Can we learn anything useful from the past e.g., from CORBA services?
- Are there research issues that need to be resolved before the picture painted here can become a reality?
- How critical is the adoption of a common set of standards for Web services?
- What is the academic perspective on Web services?
- What is the industrial perspective?
- Would it be possible to speculate on a time line for these developments?
Storage Service Providers: a Solution for Storage Management?
Thursday 13th September, 11:00-12:30, Auditorium
Chair: Banu Ozden (Bell Labs)
Panelists: Jim Gray (Microsoft Research), John Wilkes (HP Labs), and Bruce Gordon (StorageNetworks)
We are witnessing an increasing number of Storage Service Providers (SSP) emerging on the premise that SSPs can provide managed storage services more efficiently than enterprises, which are allocating a large percentage of their IT spending on storage and its management. Although there is a belief that managing large amounts of storage is difficult and expensive, the problems of storage management are not well identified and quantified, and the commercial viability of the SSP model is not assured. The goal of this panel is to discuss 1) management challenges related to data storage; 2) issues surrounding the SSP paradigm to understand under which conditions this paradigm provides a viable alternative; and 3) the implications of the SSP paradigm for database technology.
Some potential issues of controversy and questions:
- Are Storage Service Providers (SSP) here to stay?
- Can SSPs provide the same level of performance and security?
- Would you outsource your company's data?
- For what applications SSPs provide a better solution?
- What is really the cost of storage management?
- Popular number used today claims that for each $1 spent on storage equipment, $8 is spent for management. Is this correct?
- What are the storage management tasks?