Distributed Computing Class Library
|
What is DCCL DCCL is an easy to use and robust framework for distributed computing which is written in C++ for optimal memory management and efficiency. The framework consists of a single master and an arbitrary number of slaves. The master is responsible for distributing jobs to slaves and to monitor the slaves and all jobs. During development we have tried to keep the architecture as simple as possible. Since we think that failures like network errors or slave crashes are the rule and not the exception we have designed a system which guarantees that no job will be lost due to network errors, slave crashes or other failures as long as the master is up and running.  Each slave periodically pings the master to tell it that it is still alive. This mechanism also allows the automatic detection of new slaves as they are started and become available. The master also contains an http interface that can be used to monitor the current status of the master and of all slaves. For each slave some statistics are shown (e.g. load, cpu, memory usage) about the machine the slave is running on. Contact and bug reporting If you want to report bugs or have suggestions or comments feel free to send an email to the author Daniel Etzold at etzold@cs.tum.edu. |