Exploiting mean field analysis to model performances of big data architectures
详细信息 查看全文
- 作者：Aniello Castiglionea ; castiglione@acm.org" class="auth_mail ; castiglione@ieee.org" class="auth_mailAuthor Vitae ; Marco Gribaudob ; gribaudo@elet.polimi.it" class="auth_mailAuthor Vitae ; Mauro Iaconoc ; mauro.iacono@unina2.it" class="auth_mailAuthor Vitae ; Francesco Palmierid ; fpalmier@unina.it" class="auth_mailAuthor Vitae
- 关键词：Multi-formalism modeling ; Big data ; Mean field analysis ; Performance evaluation ; Map-reduce architectures
- 刊名：Future Generation Computer Systems
- 出版年：July 2014
- 年：2014
- 卷：37
- 期：Complete
- 页码：203-211
- 全文大小：959 K
文摘
- •
The paper presents a set of methods for approximate inference of probabilistic models.
- •
The proposed approach limits the excessive classical state-space growth problem.
- •
The MFA can be applied to modeling problems with very large scale stochastic systems.
- •
The MFA is effective and reliable in evaluating the performance of very large big data.
- •
The MFA is able to model performance of big data architectures indices in a bounded time.