Protocol Overview
From Unofficial BOINC Wiki
The BOINC Client Software communicates with several Servers in the course of getting work and returning Results. All communication uses HTTP on port 80, so the BOINC Client Software can function through Firewalls and Proxies.
- The BOINC Client Software downloads the page from Project's Master URL. From XML tags embedded in this page, it obtains a list of Domain Names of Schedulers.
- The BOINC Client Software exchanges request and reply messages with a Scheduling Server. The reply message contains, among other things, descriptions of work to be performed, and lists of URLs of the input and output files of that work.
- The BOINC Client Software downloads files (Science Application programs and Work Unit Data Files) from one or more download Data Servers. This uses standard HTTP GET requests, perhaps with Range commands to resume incomplete transfers.
- After the computation is complete, the BOINC Client Software uploads the Result Data Files. This uses a BOINC-specific protocol that protects against DOS Attacks on Data Servers.
- The BOINC Client Software then contacts a Scheduling Server again, reporting the completed work and requesting more work.
[edit] UCB Source
[edit] Copyright ©
- 2005 University of California
- 2005 Paul D. Buck
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
