C. R. Hofmeister and J. M. Purtilo. A framework for dynamic reconfiguration of distributed programs. Technical Report CS-TR-3119, University of Maryland, College Park, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....is one that upgrades all the nodes in the system in a timely manner, transforms their persistent state from the old version s representation to one that makes sense in the new version, and does so with minimal service disruption. Earlier approaches to automatically upgrading distributed systems [12 14, 26, 28, 38] or distributing software over networks [1 6,8,22,42] do little to ensure continuous service during upgrades. The Eternal system [45] the Simplex architecture [41] and Google [20] enable specific kinds of systems to provide service during upgrades, but they do not provide general solutions. ....

....implementations can cause inconsistent views of a node s state at di#erent versions. Section 8 elaborates on these points. 7 Transform Functions Transform functions (TFs) are procedures defined by the upgrader to convert a node s persistent state from one version to the next. In previous systems [14,19,26], TFs converted the old object into a new one whose representation (a.k.a. rep ) reflected the state of the old one at the moment the TF ran. Our system extends this approach to allow the TF to also 13 p O i f O i 1 f Initialize new SO Transform Function Before i 1 upgrade ....

[Article contains additional citation context not shown here]

C. R. Hofmeister and J. M. Purtilo. A framework for dynamic reconfiguration of distributed programs. Technical Report CS-TR-3119, University of Maryland, College Park, 1993.

....upgraded and others are not. Nonetheless, the system as a whole should continue to provide service. The goal of our research is to support automatic upgrades for such systems and to enable them to provide service during upgrades. Earlier approaches to automatically upgrading distributed systems [9 11, 17, 18, 22] or distributing software over networks [1 6,8,15,25] do little to ensure continuous service during upgrades. The Eternal system [27] the Simplex architecture [24] and Google [14] enable specific kinds of systems to provide service during upgrades, but they do not provide general solutions. An ....

....i to version i 1. Squares represent the states of the current version; circles represent the states of simulation objects. 6 Transform Functions Transform functions (TFs) are procedures defined by the upgrader to convert a node s persistent state from one version to the next. In previous systems [11, 13, 17], TFs converted the old object into a new one whose representation (a.k.a. rep ) reflected the state of the old one at the moment the TF ran. Our system extends this approach to allow the TF to also access the future SO created for its version, as illustrated in Figure 2. The TF must then produce ....

C. R. Hofmeister and J. M. Purtilo. A framework for dynamic reconfiguration of distributed programs. Technical Report CS-TR-3119, University of Maryland, College Park, 1993.

....an abstract data type. Restore initialises a compatible ADT with the state extracted by capture. Such state access and modification operations are typically restricted to use by the dynamic configuration manager [10] Persistent state management however, introduces several non trivial problems [2, 5] such as identification of critical state and transformation of state from one component to another where the data structures employed by the two components are not identical. Further, initialisation of components which are dynamically introduced may require access to state of a number of existing ....

Hofmeister, C.R. and Purtilo, J.M. A Framework for Dynamic Reconfiguration of Distributed Programs. University of Maryland, College Park, Computer Science Department. UMCP TR3119. 1993.

.... [Schiefer93] distributed systems, where the system is distributed over a network and it is undesirable to locate all components and shut them down to perform an update this is known as dynamic reconfiguration, examples of which include Conic [Kramer85] Darwin [Eisenbach93] POLYLITH [Hofmeister93] and Gerel [Endler93] languages which have dynamic linking to external program components, e.g. Napier88 [Morrison89] db Pascal [Gantenbein86] and Java) where the software components which will be bound together to form the program cannot be determined at the commencement of the program ....

....a program is a tuple #function, stack position, variable or PC#. A state mapping is defined to be a function which translates statement positions and variables in an old version to a state in the new version. Transfer of state in distributed systems has also been studied. The work of Hofmeister [Hofmeister93, Hofmeister94], deals with this issue. The approach here is similar to that of Goullon et al. where an encode and decode operation must be supplied which transforms the state into a common representation. The various approaches to state transfer described above can be categorised as those where change is ....

C. R. Hofmeister and J. M. Purtilo. A framework for dynamic reconfiguration of distributed programs. Technical report UMIACS-TR-93-78. University of Maryland Institute for Advanced Computer Studies, Department of Computer Science, University of Maryland, College Park, Maryland, 1993.

.... interface definition, component structure definition, decoupling between component functionality and configuration, type of configurationsupported according to the classification of configuration as component configuration, logical structure configuration and geometry configuration, given in [9]. Regis Darwin [1] 10] is a programming environment for the development and execution of distributed programs. The Regis approach is based on code skeleton generation out of Darwin configuration specifications. The Darwin configuration language [11] 12] represents a language support for the ....

....ffl Decouples component functionality from component static configuration. Component functionality implementation are orthogonal to configuration issues. ffl Supports static configuration of components, logical structure and geometry not supporting any kind of dynamic configuration. In [9] an extension to the work of Polylith is presented. This extension is focused on the introduction of reconfiguration capabilities to Polylith. This work adds the following characteristics. It supports all kinds of dynamic configuration. However, it does not decouple component functionality from ....

[Article contains additional citation context not shown here]

C. Hofmeister and J. Purtilo. A Framework for Dynamic Reconfiguration of Distributed Systems. In Proceedings of the 11th International Conferenceon Distributed Computing Systems, pages 560--571, 1991.

....configuration is presented as the process of modifying or extending an application without suspending its execution. This process is driven by change specifications such as, introduction of new components, modification of existing ones and modification of existing communication patterns. In [Hofmeister 91] configuration is classified into module implementation configuration, structure configuration and geometry configuration. Module implementation configuration deals with the capability of specifying and changing a component s implementation. Structure configuration deals with the capability of ....

C. Hofmeister and J. Purtilo. A Framework for Dynamic Reconfiguration of Distributed Systems. In Proceedings of the 11th International Conference on Distributed Computing Systems, pages 560--571, 1991.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC