skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: The Rendezvous Algorithm for Domain-to-Domain Data Transfers


This report describes the Rendezvous algorithm and its use in demonstrating scalability for repeated transfers and the newer implementation of the Data-Transfer Toolkit (DTK).

  1. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Publication Date:
Research Org.:
Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Sponsoring Org.:
USDOE National Nuclear Security Administration (NNSA)
OSTI Identifier:
Report Number(s):
DOE Contract Number:
Resource Type:
Technical Report
Country of Publication:
United States
42 ENGINEERING; 97 MATHEMATICS AND COMPUTING; Rendezvous algorithm; MPI; data transfer; communications

Citation Formats

Christon, Mark Allen. The Rendezvous Algorithm for Domain-to-Domain Data Transfers. United States: N. p., 2015. Web. doi:10.2172/1169154.
Christon, Mark Allen. The Rendezvous Algorithm for Domain-to-Domain Data Transfers. United States. doi:10.2172/1169154.
Christon, Mark Allen. 2015. "The Rendezvous Algorithm for Domain-to-Domain Data Transfers". United States. doi:10.2172/1169154.
title = {The Rendezvous Algorithm for Domain-to-Domain Data Transfers},
author = {Christon, Mark Allen},
abstractNote = {This report describes the Rendezvous algorithm and its use in demonstrating scalability for repeated transfers and the newer implementation of the Data-Transfer Toolkit (DTK).},
doi = {10.2172/1169154},
journal = {},
number = ,
volume = ,
place = {United States},
year = 2015,
month = 2

Technical Report:

Save / Share:
  • Scientific applications and experimental facilities generate massive data sets that need to be transferred to remote collaborating sites for sharing, processing, and long term storage. In order to support increasingly data-intensive science, next generation research networks have been deployed to provide high-speed on-demand data access between collaborating institutions. In this paper, we present a practical model for online data scheduling in which data movement operations are scheduled in advance for end-to-end high performance transfers. In our model, data scheduler interacts with reservation managers and data transfer nodes in order to reserve available bandwidth to guarantee completion of jobs that aremore » accepted and confirmed to satisfy preferred time constraint given by the user. Our methodology improves current systems by allowing researchers and higher level meta-schedulers to use data placement as a service where theycan plan ahead and reserve the scheduler time in advance for their data movement operations. We have implemented our algorithm and examined possible techniques for incorporation into current reservation frameworks. Performance measurements confirm that the proposed algorithm is efficient and scalable.« less
  • The future of extreme-scale computing is expected to magnify the influence of soft faults as a source of inaccuracy or failure in solutions obtained from distributed parallel computations. The development of resilient computational tools represents an essential recourse for understanding the best methods for absorbing the impacts of soft faults without sacrificing solution accuracy. The Rexsss (Resilient Extreme Scale Scientific Simulations) project pursues the development of fault resilient algorithms for solving partial differential equations (PDEs) on distributed systems. Performance analyses of current algorithm implementations assist in the identification of runtime inefficiencies.
  • A numerical algorithm is described for computing coupled heat and hydrodynamic flow on an arbitrarily shaped, time-varying domain. The algorithm extends the ICED-ALE method, for the computation of hydrodynamic motion using an arbitrary Eulerian-Lagrangian mesh, to include electron thermal conduction. Appropriate difference equations in conservation form are developed, and the Dufort-Frankel scheme is adapted to advance these equations in time. Results are presented which illustrate the application of the algorithm to the two-dimensional hydrodynamic motion induced by illuminating spherical and thin foil targets with a single laser beam.
  • AN algorithm which efficiently solves large systems of equations arising from the discretization of a single second-order elliptic partial differential equation is discussed. The global domain is partitioned into not necessarily disjoint subdomains which are traversed using the Schwarz Alternating Procedure. On each subdomain the multigrid method is used to advance the solution. The algorithm has the potential to decrease solution time when data is stored across multiple levels of a memory hierarchy. Results are presented for a virtual memory, vector multiprocessor architecture. A study of choice of inner iteration procedure and subdomain overlap is presented for a model problem,more » solved with two and four subdomains, sequentially and in parallel. Microtasking multiprocessing results are reported for multigrid on the Alliant FX-8 vector-multiprocessor. A convergence proof for a class of matrix splittings for the two-dimensional Helmholtz equation is given. 70 refs., 3 figs., 20 tabs.« less
  • A current area of interest in finite-difference time-domain computer is in developing the ability to model items of less than one cell in extent. A special case of this is an attempt to provide an algorithm to model thin slots, (much less than one cell dimension), without resorting to small cells which would increase storage and time requirements. Such an algorithm has been written by Yee and Kasher, and seems to work well. However, to lend real confidence to the algorithm's effectiveness, comparison with experimental data was deemed necessary. To this end, a series of experiments were carried out inmore » the Lawrence Livermore National Laboratory's EMPEROR facility. This report will describe the measurements, the facility in which they were taken, and will conclude with a comparison of the experimental data and the results obtained from a finite-difference code with the thin slot algorithm. The comparison show that the thin slot algorithm performs as well or better than expected, with generally less than a factor of 2 difference in amplitude. 6 refs., l34 figs.« less