[Updated on Apr. 18] Integration of CiNii Articles into CiNii Research


Bibliographic Information

Other Title
  • Performance Optimization Based on Exchange of Computational Resources between Container-based Virtual Environment

Search this article



Many programs have space-time trade-offs: e.g. programs which cache computation results on memory or programs use garbage collection can improve their throughput with more memory space. These programs may modify computational resource allocation without changing their performance, by increasing the memory usage instead of reducing the CPU usage, for example, or vice versa. By utilizing this property, in the case of multiple instances of such programs, the performance of at least one such program can be improved by trading CPU utilization and memory space between programs. In recent years, cloud computing services are widely used. Cloud provider creates isolated environments in servers using some virtualization technology and is running various programs on those environments. By applying this idea here, we can expect improvement in terms of utilization efficiency of the servers. In this presentation, we propose a system for optimizing computational resource allocation automatically with exchange of computational resources between Linux containers which running programs with space-time trade-offs including Key-Value Store for accelerate computation.




Report a problem

Back to top