SIMUTools2011

Discrete Stochastic Simulation of Spatially Inhomogeneous Biochemical Systems

More about Linda...

Linda Petzold - Professor, Departments of Computer Science and Mechanical Engineering University of California, Santa Barbara, California, U.S.A.

Abstract: In microscopic systems formed by living cells, the small numbers of some reactant molecules can result in dynamical behavior that is discrete and stochastic rather than continuous and deterministic. An analysis tool that respects these dynamical characteristics is the stochastic simulation algorithm (SSA), which applies to well-stirred chemically reacting systems. However, cells are hardly homogeneous! Spatio-temporal gradients and patterns play an important role in many biochemical processes. In this lecture we report on recent progress in the development of methods for spatial stochastic and multiscale simulation, and outline some of the many interesting complications that arise in the modeling and simulation of spatially inhomogeneous biochemical systems.

Predicting Performance of Large Scale Web Search Engines using Discrete-Event Simulation

More about Ricardo...

Ricardo Baeza-Yates - VP of Yahoo! Research for Europe and Latin America

Abstract: Dealing efficiently with multiple user queries, each at a different stage of execution at any given instant of time, is a key issue in large-scale Web search engines. Here the use of suitable parallel computing techniques devised to achieve a high query throughput upon the least possible amount of hardware, and yet remain stable under sudden peaks in traffic, is critical to efficient data center operation. Achieving this goal is quite beyond the possibilities of a single indexing data structure and respective query processing algorithm. Indeed, current practice in Web search engines clearly indicates that such a goal is only feasible through a combination of algorithms and heuristics specifically devised to work together so that they -- as a whole -- lead to efficient and scalable performance upon well-dimensioned hardware. This in turn makes performance prediction of these complex systems mathematically intractable.

In this talk a technique for performance prediction based on discrete-event simulation is described and illustrated with examples from power consumption prediction and capacity planning for data centers. Salient features of the technique are (i) the use of models of parallel computing to model and cost hardware resources, (ii) the use of process oriented discrete-event simulation to model and cost the effects of multiple queries competing for using hardware resources, (iii) reduction of the number of simulations required to find an optimal deployment of search engine components onto cluster nodes by using mean-value analysis and novel cost-formula, and (iv) execution of simulations driven by actual user query logs and cost of primitive query solution operations.