AUTOMATEENABLING AUTONOMIC GRID APPLICATIONS: REQUIREMENTS, MODELS AND INFRASTRUCTUREPRESENTER: Khalid SaleemAGNOSTIC: Fernando FarfánAGNOSTIC QUESTIONS1. How does AutoMate deal with the heterogeneity among all theelements of a grid? How does AutoMate integrate so manyheterogeneous autonomic components to work together? By separating the policy from mechanism. Policies inthe form of rules are used to orchestrate a repertoireof mechanisms to achieve context-aware adaptiveruntime computational behaviors and coordination andinteraction relationships based on functional,performance, and quality of service requirements thusresponding to heterogeneity and dynamics. UsesDynamic composition to enable relationships betweenelements (via element managers) to be establishedand modified at runtime.2. Security and trust is one of the key issues in grid environments,and the maintenance of users and permissions may become abottleneck. Does AutoMate provide a specific autonomic solutionfor a self-maintenance of users and permissions acrossadministrative boundaries? AutoMate does not explicitly provide a methodologyfor self maintenance of users and permissions acrossadministrative boundaries. However, SESAME that isembedded into AutoMate allows for setting up securitypolicies and permissions for avoiding fraud andintrusion. 3. The Accord framework incorporates practical human knowledgein the form of behavioral rules, but with the vastness ofcommunication paradigms (e.g., RPC, RMI, publish/subscribe),how does Accord ensure the correctness of these rules? “To ensure rule correctness, we design a rule templatefor each communication paradigm (e.g., RPC/RMI,publish/subscribe) and coordination pattern (e.g.,conditional branch, loop, sequence, parallel execution).A rule template is instantiated via filling inparameters, such as the names of interaction partiesand the exchanging data.”http://www.caip.rutgers.edu/~marialiu/Projects/Accord/index.htm 4. AutoMate utilizes a peer-to-peer middleware to support andenable the autonomic interactions among components. Can youthink of a better architecture to achieve this goal? In this scenario using peer-to-peer would be a betteroption as AutoMate is based upon autonomic agentbased control networks. Element managers executerules to establish control and communicationrelationships among these elements in a decentralizedand parallel manner.5. Does Accord target a specific programming language? Nothing particular. However, the prototypeimplementation was done using C++ and MPI6. Can you extend on how AutoMate lets Globus-enabled grids toachieve an autonomic management? What elements of theAutoMate architecture communicate with Globus elements? AutoMate makes use of its Accord autonomic elementsand rule set, the Rudder Agent Framework and COMETto provide dynamic flexible and scaleable coordinationamong the peer nodes along with the use of conflictresolution and negotiation protocols. No specificinformation mapping the components of AutoMate tothat of Globus has been provided. We can find a mapping between GT4 components andAutoMate components, but this is not clearly stated.7. Are you aware of other projects with the same goals thanAutoMate? Grid MP by United Devices http://www.ud.com/products/gridmp_fabs.php Optimal Gridhttp://www-128.ibm.com/developerworks/library/gr-opgrid/ GridARM for Structured Adaptive Mesh Refinement (SAMR) application  OGSA to some extent8. On a previous presentation of the same authors, we discussedthat their so-called peer-to-peer architecture was in fact Client-Server. Do you think it is correct that they somewhat reuse thisarchitecture for the Meteor and COMET components? The previous presentation focused on PAWN where asthe authors made use of Meteor in the currentframework. Meteor is based on JXTA which is a generalpurpose P2P framework. Hence, it seems correct to useMeteor.ADDITIONAL DISCUSSION The very most important point of the paper is missing.They don’t make clear what is the need of autonomic computing inthe presented grid examples. The authors do not specify how the intercommunicationbetween programming elements happens. There is an overlap between the operational port andthe control port. Operational port is supposed to do some of thework of the Control port. In the Forest Fire Management simulation, there is nonotion of grid computing. The use of grid computing is missing. The general feeling abut this paper is that the authorsare just packing a lot of related concepts just to sell it to adifferent