5 <img align=center src="simgrid_logo.png" alt="SimGrid"><br>
9 \section overview Overview
11 SimGrid is a toolkit that provides core functionalities for the simulation
12 of distributed applications in heterogeneous distributed environments.
13 The specific goal of the project is to facilitate research in the area of
14 distributed and parallel application scheduling on distributed computing
15 platforms ranging from simple network of workstations to Computational
18 \section people People
20 The authors of SimGrid are:
22 \author Henri Casanova <casanova@cs.ucsd.edu>
23 \author Arnaud Legrand <arnaud.legrand@imag.fr>
24 \author Martin Quinson <martin.quinson@tuxfamily.org>
26 \section intro Available Softwares
28 The SimGrid toolkit is composed of different modules :
30 \li XBT (eXtensive Bundle of Tools) is a portable library with many
31 convenient portable datastructures (vectors, hashtables, heap,
32 contexts ...). Most other SimGrid modules rely on it.
34 \li SURF provides the core functionnalities to simulate a virtual
35 platform. It is very low-level and is not intended to be used as
36 such but rather to serve as a basis for higher-level simulators
37 (like MSG, GRAS, SMPI, ...). It relies on a fast max min linear
40 \li MSG is a simulator built using the previous modules. It aims at
41 being realistic and is application-oriented. It is the software layer
42 of choice for building simulation with multiple scheduling agents.
44 \li GRAS (<em>not functionnal yet</em>) is an ongoing project to emulate virtual
45 virtual platforms through SURF. As a consequence a code developped using the GRAS
46 framework is able to run as well in the real-world as in the
47 simulator. The resulting code is very portable and highly interoperable while
48 remaining very efficient. Even if you do not plan to run your code for real,
49 you may want to switch to GRAS if you intend to use MSG in a very intensive way
50 (e.g. for simulating a peer-to-peer environment).
52 \li SMPI (<em>not functionnal yet</em>) is an ongoing project to enable MPI code
53 to run on top of a virtual platform through SURF. It follows the same principle as
54 the ones used in GRAS but is specific to MPI applications.
56 Here is a figure the depicts the relation between those different modules.
59 <img align=center src="simgrid_modules.jpg" alt="SimGrid"><br>
63 The section \ref publications contains links to papers that provide
64 additional details on the project as well as validation and
67 The software can be downloaded from <a href="http://gcl.ucsd.edu/simgrid/dl/">here</a>.
69 \section install Installation
72 \li <tt>./configure</tt>
73 \li <tt>make all install</tt>
75 If you are not familiar with compiling C files under UNIX and using
76 libraries, you will find some more informations in Section \ref
79 \section documentation API Documentation
81 The API of all different modules is described <a href="API/html/modules.html">here</a>.
82 See <a href="examples/html/modules.html">here</a> for an introduction on the way to use these modules.
84 \section users_contributers Users / Contributers
86 \subsection contributers Contributers
88 \li Loris Marchal: wrote the new algorithm for simulation TCP
90 \li Julien Lerouge : wrote a XML parser for ENV descriptions and
91 helped for the general design during a 4 month period (march-june 2002)
93 \li Clément Menier and Marc Perache : wrote a first prototype of
94 the MSG interface during a project at ENS-Lyon (jan 2002).
95 \li Dmitrii Zagorodnov : wrote some parts of the first version
98 \subsection mailinglist User Mailing List
100 We have a <a href=https://listes.ens-lyon.fr/wws/info/simgrid2-users> mailing list for
101 SimGrid users</a>.<p>
103 \section publications Publications
105 \subsection simulation About simulation
107 \li <b>Scheduling Distributed Applications: the
108 SimGrid Simulation Framework</b>\n
109 by <em>Henri Casanova and Arnaud Legrand and Loris Marchal</em>\n
110 Proceedings of the third IEEE International Symposium
111 on Cluster Computing and the Grid (CCGrid'03)\n
112 Since the advent of distributed computer systems an active field
113 of research has been the investigation of scheduling strategies
114 for parallel applications. The common approach is to employ
115 scheduling heuristics that approximate an optimal
116 schedule. Unfortunately, it is often impossible to obtain
117 analytical results to compare the efficacy of these heuristics.
118 One possibility is to conducts large numbers of back-to-back
119 experiments on real platforms. While this is possible on
120 tightly-coupled platforms, it is infeasible on modern distributed
121 platforms (i.e. Grids) as it is labor-intensive and does not
122 enable repeatable results. The solution is to resort to
123 simulations. Simulations not only enables repeatable results but
124 also make it possible to explore wide ranges of platform and
125 application scenarios.\n
126 In this paper we present the SimGrid framework which enables the
127 simulation of distributed applications in distributed computing
128 environments for the specific purpose of developing and evaluating
129 scheduling algorithms. This paper focuses on SimGrid v2, which
130 greatly improves on the first version of the software with more
131 realistic network models and topologies. SimGrid v2 also enables
132 the simulation of distributed scheduling agents, which has become
133 critical for current scheduling research in large-scale platforms.
134 After describing and validating these features, we present a case
135 study by which we demonstrate the usefulness of SimGrid for
136 conducting scheduling research.
139 \li <b>A Network Model for Simulation of Grid Application</b>\n
140 by <em>Henri Casanova and Loris Marchal</em>\n
142 In this work we investigate network models that can be
143 potentially employed in the simulation of scheduling algorithms for
144 distributed computing applications. We seek to develop a model of TCP
145 communication which is both high-level and realistic. Previous research
146 works show that accurate and global modeling of wide-area networks, such
147 as the Internet, faces a number of challenging issues. However, some
148 global models of fairness and bandwidth-sharing exist, and can be link
149 withthe behavior of TCP. Using both previous results and simulation (with
150 NS), we attempt to understand the macroscopic behavior of
151 TCP communications. We then propose a global model of the network for the
152 Grid platform. We perform partial validation of this model in
153 simulation. The model leads to an algorithm for computing
154 bandwidth-sharing. This algorithm can then be implemented as part of Grid
155 application simulations. We provide such an implementation for the
156 SimGrid simulation toolkit.\n
157 ftp://ftp.ens-lyon.fr/pub/LIP/Rapports/RR/RR2002/RR2002-40.ps.gz
160 \li <b>MetaSimGrid : Towards realistic scheduling simulation of
161 distributed applications</b>\n
162 by <em>Arnaud Legrand and Julien Lerouge</em>\n
163 Most scheduling problems are already hard on homogeneous
164 platforms, they become quite intractable in an heterogeneous
165 framework such as a metacomputing grid. In the best cases, a
166 guaranteed heuristic can be found, but most of the time, it is
167 not possible. Real experiments or simulations are often
168 involved to test or to compare heuristics. However, on a
169 distributed heterogeneous platform, such experiments are
170 technically difficult to drive, because of the genuine
171 instability of the platform. It is almost impossible to
172 guarantee that a platform which is not dedicated to the
173 experiment, will remain exactly the same between two tests,
174 thereby forbidding any meaningful comparison. Simulations are
175 then used to replace real experiments, so as to ensure the
176 reproducibility of measured data. A key issue is the
177 possibility to run the simulations against a realistic
178 environment. The main idea of trace-based simulation is to
179 record the platform parameters today, and to simulate the
180 algorithms tomorrow, against the recorded data: even though it
181 is not the current load of the platform, it is realistic,
182 because it represents a fair summary of what happened
183 previously. A good example of a trace-based simulation tool is
184 SimGrid, a toolkit providing a set of core abstractions and
185 functionalities that can be used to easily build simulators for
186 specific application domains and/or computing environment
187 topologies. Nevertheless, SimGrid lacks a number of convenient
188 features to craft simulations of a distributed application
189 where scheduling decisions are not taken by a single
190 process. Furthermore, modeling a complex platform by hand is
191 fastidious for a few hosts and is almost impossible for a real
192 grid. This report is a survey on simulation for scheduling
193 evaluation purposes and present MetaSimGrid, a simulator built
195 ftp://ftp.ens-lyon.fr/pub/LIP/Rapports/RR/RR2002/RR2002-28.ps.gz
197 \li <b>SimGrid: A Toolkit for the Simulation of Application
199 by <em>Henri Casanova</em>\n
200 Advances in hardware and software technologies have made it
201 possible to deploy parallel applications over increasingly large
202 sets of distributed resources. Consequently, the study of
203 scheduling algorithms for such applications has been an active area
204 of research. Given the nature of most scheduling problems one must
205 resort to simulation to effectively evaluate and compare their
206 efficacy over a wide range of scenarios. It has thus become
207 necessary to simulate those algorithms for increasingly complex
208 distributed, dynamic, heterogeneous environments. In this paper we
209 present SimGrid, a simulation toolkit for the study of scheduling
210 algorithms for distributed application. This paper gives the main
211 concepts and models behind SimGrid, describes its API and
212 highlights current implementation issues. We also give some
213 experimental results and describe work that builds on SimGrid's
215 http://grail.sdsc.edu/papers/simgrid_ccgrid01.ps.gz
217 \subsection research Papers using SimGrid results
219 \li <b>Optimal algorithms for scheduling divisible workloads on
220 heterogeneous systems</b>\n
221 by <em>Olivier Beaumont and Arnaud Legrand and Yves Robert</em>\n
222 In this paper, we discuss several algorithms for scheduling
223 divisible loads on heterogeneous systems. Our main contributions
224 are (i) new optimality results for single-round algorithms and (ii)
225 the design of an asymptotically optimal multi-round algorithm. This
226 multi-round algorithm automatically performs resource selection, a
227 difficult task that was previously left to the user. Because it is
228 periodic, it is simpler to implement, and more robust to changes in
229 the speeds of processors or communication links. On the theoretical
230 side, to the best of our knowledge, this is the first published
231 result assessing the absolute performance of a multi-round
232 algorithm. On the practical side, extensive simulations reveal
233 that our multi-round algorithm outperforms existing solutions on a
234 large variety of platforms, especially when the
235 communication-to-computation ratio is not very high (the difficult
237 ftp://ftp.ens-lyon.fr/pub/LIP/Rapports/RR/RR2002/RR2002-36.ps.gz
238 \li <b>On-line Parallel Tomography</b>\n
239 by <em>Shava Smallen</em>\n
240 Masters Thesis, UCSD, May 2001
241 \li <b>Applying Scheduling and Tuning to On-line Parallel Tomography </b>\n
242 by <em>Shava Smallen, Henri Casanova, Francine Berman</em>\n
243 in Proceedings of Supercomputing 2001
244 \li <b>Heuristics for Scheduling Parameter Sweep applications in
245 Grid environments</b>\n
246 by <em>Henri Casanova, Arnaud Legrand, Dmitrii Zagorodnov and
247 Francine Berman</em>\n
248 in Proceedings of the 9th Heterogeneous Computing workshop
249 (HCW'2000), pp349-363.