powered and hosted by
University of Federal Armed Forces,
Institute for information Systems
Munich, Germany














Panels










Five Panel discussions will be held during IM 2007.


Tuesday, 22 May 2007

13:30 - 15:10

PS1 - Techniques for Policy Refinement: The Importance of Going One Level Down!

Chair: Jorge Lobo, IBM T.J. Watson, USA

Policy Refinement refers to the process of deriving concrete implementable policies from higher-level goals or Service Level Agreements. It has been presented often as one of the most desirable research objectives in policy-based management yet one of the most challenging to address. In recent years, several approaches that have made some headway towards addressing this goal have emerged. They include amongst others: goal elaboration and abductive reasoning, model checking, case-based reasoning, and model transformation based on models and ontologies. The aim of this panel is to confront these approaches and in the process a number of questions will undoubtedly arise: are any of these approaches viable? How can refined policies be analysed e.g., for conflicts? How much user intervention is required? When is each approach applicable? Can these approaches be combined?

Panelists:
  • Emil Lupu, Imperial College London, Great Britain
  • Joan Serrat, Universitat Politecnica de Catalunya, Spain
  • Seraphin Calo, IBM T.J. Watson Research, USA
  • John Strassner, Motorola Research Labs, USA


Wednesday, 23 May 2007

10:30 - 12:10

PS2 - The End of the Waterfall -- Integrating Application Development and Operational Management

Chair: Tamar Eilam, IBM T.J. Watson Research, USA

Two trends are pushing the industry towards the merger of application development and operational management, activities that have largely been treated as distinct phases in the software lifecycle.

The first trend, which has been on-going for some time, is that it is very difficult to set up meaningful test environments for distributed applications, with test setup consuming a large fraction of the test time. These difficulties here are a consequence of the need to: (1) have a large scale environment in which to test and (2) access realistic data in order to conduct meaningful tests, and (3) have multiple different test environments for multiple test phases.

The second trend is architectural styles such as SOA, mesh-ups, and Web2.0 in which programmers integrate services from live web sites. Such composites of running services are a very different style of development and deployment from the use of programming frameworks such as J2EE and .NET.

There are several questions that arise from the integration of application development and operational management:

  1. How can operational services be constructed and used to facilitate application debugging and testing (e.g., services introspection, on-line help, isolation of test and operational users)?
  2. What extensions are needed to applications and application development tools (e.g., ECLIPSE, Visual Studio) in order to support development, debugging, and management in operational settings?
  3. How can knowledge of the data center structure, capabilities and constraints, can be used to develop applications optimized for operation in a given data center?
  4. How should non-functional requirements such as security and availability be used as input throughout the software life cycle?
  5. What additional unique opportunities emerge in merging development and operation? For example, can this merger result in a dramatic acceleration of the iterative programming style advocated by the Rational Unified Process (RUP)?
Panelists:
  • Joe Hellerstein, Microsoft, USA
  • German Goldszmidt, IBM, USA
  • Jerry Rolia, HP Laboratories, USA
  • Mark Burgess, University College Oslo, Norway


13:30 - 15:10

PS3 - Can Management Systems be Trusted?

Chair: Hanan Lutfiyya, University of Western Ontario, Canada

There is currently a good deal of discussion on the development of autonomic management systems. A question often asked is whether or not IT Managers will trust autonomic management systems. The question should be extended to ask about the level of trust that IT managers have in current management systems. Many IT managers feel that the decisions made by the management system would not be as good as the one they would make. Is there
enough of a lack of trust in current management systems that would make it difficult for to IT managers to trust and thus use autonomic management systems? Questions include the following:

  • What is the perception that IT managers have of the accuracy of management systems? What do IT managers trust e.g., Do IT managers e even trust that monitoring is accurate? What sorts of decisions do IT managers trust? How does performance impact trust?
  • Perceptions are often different than reality. What should IT managers trust? What should IT managers not trust?
  • How much impact does a lack of trust have in going to the next level of automation?
  • What features must management systems provide to increase trust?
  • How do we measure trust? How should we measure trust?
Panelists:
  • Ehab El-Shaer, DePaul University, USA
  • Alexander Keller, IBM, USA - download presentation (pdf)
  • Hong Li, Intel, USA
  • Jacques Sauve, Federal University of Campina Grande, Brazil


Thursday, 24 May 2007

10:30 - 12:10

PS4 - Virtual Machine Management: Old Wolf in New Sheep's Clothing?

Chair: Omar Cherkaoui, Universite du Quebec a Montreal, Canada

Virtualization allows isolation between applications and portability of applications over OS and hardware resources.  But another main challenge of virtualization is to allow the decoupling of infrastructure providers (who deploy and maintain network equipment) from service providers (who deploy network protocols and offer end-to-end services). This new network virtualization will help the deployment of shared experimental facilities, such as PlanetLab and GENI.  Those projects try to build this virtual network that will consist of virtual nodes and links that belong to the same service provider. This panel will try to give an answer to the mains issues behind this new virtual network:

  • How can we use virtualization technology (XEN, VMware, etc.) to build virtual nodes?
  • Why is an old technology like virtual machines suddenly so important again?
  • For what reason to people deploy virtualization technology?
  • How do people deal with the increased number of machines that need to be managed, and patched?
  • After virtualizing networks, machines, operating systems,..., what is the next thing to virtualize?
  • What are the network management functions we need to build and to deploy this new virtual network?
  • What are the issues for the network management community to solve the migration and configuration of those virtual nodes?
Panelists:
  • Guy Pujolle, University Paris 6, France
  • Masum Hasan, Cisco Systems, USA
  • Giovanni Pacifici, IBM, USA
  • Rene Schmidt, VMware, USA


13:30 - 14:50

PS5 - What's IT management worth to the business?

Chair: Claudio Bartolini, HP Laboratories, USA

Enterprises in every industry sector are increasingly demanding that their investment in IT is justified by keeping IT service delivery and support aligned and in synch with their business objectives. Our research community is also growing more aware of the issue, as testifies the theme of this 2007 edition of IM: "moving from bits to business value".

Our panelists will address the following questions and more from the audience.
  • What is the business value of managing networks and IT systems?
  • How can an enterprise capture this value, or even how can line of business managers be convinced that it pays off in the long run to spend money in management software?
  • What's automation really worth?
  • Would an IT manager make the same decisions as he/she makes every day if they had better visibility on the business that IT supports?
  • How is the network and system management community addressing these issues?
Panelists:
  • Mark Burgess, Oslo University College, Norway
  • Jacques Sauve, Federal University of Campina Grande, Brazil
  • John Strassner, Motorola Research Labs, USA
  • John Wilkes, HP Laboratories, USA