Section: Software
Prototype Software
Mascopt and openGVE (http://www-sop.inria.fr/mascotte/mascopt/ )
Participants : Fabrice Peix, Michel Syska.
Mascopt [122] is a free Java library distributed under the terms of the LGPL license which is dedicated to graph and network processing. Mascopt includes a collection of Java interfaces and classes that implement fundamental data structures and algorithms. The forthcoming public distribution of Mascopt will appear in january 2010 under the name of the openGVE project, Mascopt being one implementation of the bridge graph interface [R. Correa, http://opengve.inria.fr/bridge-graph-interface/apidocs/fr/inria/opengve/bridge/interfaces/Graph.html ). The objective is to allow easy integration of different implementations. The applications already written will not be affected, they will not have to be rewritten but will have different choices of internal implementation. This may lead to better performances for specific issues such as large graphs processing.
The main objective of Mascopt (Mascotte Optimization) project is to ease software development in the field of network optimization. Examples of problems include routing, grooming, survivability, and virtual network design. Mascopt helps implementing a solution to such problems by providing a data model of the network and the demands, classes to handle data and ready to use implementation of existing algorithms or linear programs (e.g. shortest paths or integral multicommodity flow).
A generic linear programming interface allows users to program the same way whether the target solver is IBM ILOG CPLEX, GLPK (GNU Linear Programming Kit) or CLP/CBC (accessed through JNI).
Mascopt has intensively been used within Mascotte industrial cooperation programs for experimentation and validation purposes: with Alcatel Space Technologies on the design of fault-tolerant on-board network satellites, on the optimization of the access layer and planning of satellite communication and with Orange Labs on the design of telecommunication backbone networks.
Another cooperation at INRIA Sophia Antipolis Méditerranée is the use of Mascopt by the Aoste team.
OSA: an Open Component-based Architecture for Discrete-Event Simulations. (http://osa.inria.fr/wiki/ )
Participants : Olivier Dalle, Fabrice Peix, Judicael Ribault.
Component-based modeling has many well-known good properties. One of these properties is the ability to distribute the modeling effort amongst several experts, each having his/her own area of system expertise. Clearly, the less experts have to care about areas of expertise of others, the more efficient they are in modeling sub-systems in their own area. Furthermore, the process of studying complex systems using discrete-event computer simulations involves several areas of non-system expertise, such as discrete-event techniques or experiment planning.
The Open Simulation Architecture (OSA) [121] is designed to enforce a strong separation of the end-user roles and therefore, ensure a successful cooperation of all the experts involved in the process of simulating complex systems.
The OSA architecture is also intended to meet the expectations of a large part of the discrete-event simulation community: it provides an open platform intended to support researchers in a wide range of their simulation activities, and allows the reuse and sharing of system models in the simulation community by means of a flexible and generic component model (Fractal).
Many discrete-event simulators are developed concurrently, but with identical or similar purpose. Another goal of OSA is to favor the reuse and integration of simulation software components and models. To favor reuse, OSA uses a layered approach to combine the modeling, simulation, and related concerns, such as instrumentation or deployment. This ability is demonstrated by the successful integration and reuse of third-party components, such as Scave, the analysis module of Omnet++, or a large number of the James II plugins developped by the University of Rostock. OSA is both a testbed for experimenting new simulation techniques and a tool for real case studies.
OSA is Open Source (LGPL) and is available for download on the INRIA forge server http://osa.gforge.inria.fr/ .
Dipergrafs (http://www-sop.inria.fr/members/Luc.Hogie/dipergrafs/ )
Participants : Luc Hogie, Issam Tahiri.
The Dipergrafs project proposes a Java framework for the manipulation of directed hypergraphs. Briefly, a directed hypergraph consists in a set of directed links, each link connecting a set of vertices to another set of vertices. In other words, a directed hypergraph is a graph in P(E). Hypergraphs are used into the fields of network modeling, rational databases, semantic web, expert systems, route planning. In particular, the design objectives of Dipergrafs are to make it particularly useful in the context of network simulation.
Briefly Dipergrafs: has a vertex-oriented design (in opposite to node-oriented design), that is the graph is seen as a collection of relations between nodes; imposes no constraint on the type of nodes and vertices (in opposite to frameworks which oblige to follow a certain structure, leading tp a lack of flexibility); provides implementations for common graph operations : navigation (paths, connected components, shortest paths, hop-exploring, etc), graph queries, graph metrics (radius, density, degrees, distance/adjacency/incidence matrices, etc), distributions of vertex metrics; is mostly usable through a small set of Java classes (in opposite to frameworks whose utilization requires the knowledge of numerous classes); features graph input/output mechanisms, allowing persistence, serialization, etc; does not feature any graph rendering tool. Instead bridges to external products dedicated to rendering are provided; comes with a set of composeable topology generator allowing to quickly instantiate the desired topology.
Dipergrafs is extensively used in the DRMSim project, in which it enables the modeling and simulation of large backbone networks.
DRMSim: (http://www-sop.inria.fr/mascotte/projets/DCR/ )
Participants : Luc Hogie, Issam Tahiri, David Coudert.
The expansion of the Internet routing system results in a number of research challenges, in particular, the Border Gateway Protocol (BGP) starts to show its limits amongst others in terms of the number of routing table entries it can dynamically process and control. Dynamic routing protocols showing better scaling properties are thus under investigation. However, because deploying under-development routing protocols on the Internet is not practicable at a large-scale (due to the size of the Internet topology), simulation is an unavoidable step to validate the properties of a newly proposed routing scheme. Unfortunately, the simulation of inter-domain routing protocols over large networks (order of tens of thousands of nodes) poses real challenges due to the limited memory and computational power that computers impose. Existing simulation tools exhibit limitations in terms of the number of nodes they can handle and in the models they propose. This motivated us for conceiving and developing an adequate network simulator call DRMSim (Dynamic Routing Model simulator) which addresses the specific problem of large-scale simulations of (inter-domain) routing models on large networks.
DRMSim relies on a discrete-event simulation engine. It proposes a general routing model which accommodates any network configuration. Aside to this, it includes specific models for GLP, and K-chordal network topologies, as well as implementations of routing protocols, including the NSR routing protocol and lightweight versions of BGP. More features will be further incorporated into the simulator. In particular, they address the challenge of simulation of larger networks (order of 10k nodes), the next step is to propose new routing algorithms, including state-of-the-art ones, to enhance the code, and to go further with distributed simulation campaigns.
DRMSim is developped in cooperation with LaBRI (Laboratoire Bordelais de Recherche en Informatique, Bordeaux, France).
