Information Technology Reference
In-Depth Information
Fig. 3.
Description of the OurGrid architecture
3.3 Volunteer Computing
The OurGrid middleware provides support for implementing the volunteer com-
puting paradigm[2,17]. Volunteer computing is based on individual users making
available their computing resources to execute applications and projects. The
first volunteer computing projects were proposed in the mid-1990s. In 1999, the
well-known
SETI@home
and
Folding@home
distributed projects were launched.
Both became pioneering projects to demonstrate the power of gathering volun-
teer computing resources to solve very large scale scientific problems.
The main features of volunteer computing are:
i) often unaccountable and potentially anonymous users voluntarily provide
their resources for computation;
ii) users can join and leave the volunteer computing platform at any time;
iii) users “credits” are accounted in order to know how much computing time
h in the systems been used and provided by every user;
iv) replication is usually applied for fault-tolerance, in order to cope with incor-
rect results or anomalous events that occur when volunteers unexpectedly
leave the system.
The middleware for volunteer computing is a software layer that provides
support for creating, managing, and using the volunteer distributed computing
infrastructure, independently from the scientific computing applications to exe-
cute. The Berkeley Open Infrastructure for Network Computing (BOINC) [1] is
the most widely used middleware system for volunteer computing. The general
architecture of a middleware for volunteer computing is like the one presented for
OurGrid: a client program runs on the volunteer side, which periodically contacts
servers over the Internet, requesting jobs and reporting the results of completed
jobs. In OurGrid, the standard volunteer computing model is extended to sup-
port a full P2P architecture. This feature allows OurGrid to be used in cloud
infrastructures too.
