PUMA
Istituto di Scienza e Tecnologie dell'Informazione     
Ciompi P., La M., Lissoni C., Martin I., Simoncini L. A proposal for a highly available multimicroprocessor system. R-82008/P/E, dicembre 1982. Document n. IEI-S82-08, 1982.
 
 
Abstract
(English)
High availability is one of the main requirements of a system in many applica- tions: process control, telecommunications, realtime processing. In the frame of the Convention between the IEI Italian National Council of Research and Selenia, a research is under development in Pisa, with the aim of implementing a highly available multimicroprocessor system. This research is based on the M.A.R.A. (Modular Architecture for Real Time Processing) multimicroprocessor architecture developed by Selenia. M.A.R.A. is intended to serve for several applications and therefore high availability policies are to be studied as an optional feature to be implemented only for those applications which require it. A straightforward way to meet this requirement is the implementation of a single high available M.A.R.A. node, using the hardware of two M.A.R.A. nodes, with a duplex organization and through a managing of the redundancy in an appli- cation transparent way. The two M.A.R.A. nodes have identical configuration and are connected to the same peripheral units. At every time, one of the two nodes (Master Node) is active and communicates with the peripheral units, while the other (Slave Node) is always ready to substitute the Master Node in the case of a fault. The Master Node communicates with the Slave Node for maintaining all the information necessary to constitute an updated and consistent copy of the system state. The two nodes have special hardware for L fault detection. The detection of a fault on the Master Node causes an immediate switching between the Master Node and the Slave Node. Periodically, a programmed switch is made, which allows the detection of faults detectable only when a node is working as Master Node. The application software is subdivided into atomic transactions". An atomic transaction is intended as a set of actions which update the system state so that either all the actions will have been executed, or none will have been. The management of fault conditions and the switching are made by the system software, so that they are transparent to the application programmer.
Subject


Icona documento 1) Download Document PDF


Icona documento Open access Icona documento Restricted Icona documento Private

 


Per ulteriori informazioni, contattare: Librarian http://puma.isti.cnr.it

Valid HTML 4.0 Transitional