Scholarlink

In this paper, we newly propose the Purpose and Read-data-from (RDF) (RP-) serializability to provide energy-efficient application services. In the PR-serializability, transactions are serialized with the significant order of purposes and the RDF relation of transactions is maintained. The Energy Efficient Purpose and RDF (EEPR) scheduler is proposed to realize the PR-serializability by improving the Energy Efficient Purpose Ordering (EEPO) scheduler proposed in our previous studies. In the EEPR scheduler, each read method can be performed on each object without blocking by any other method reading the latest committed version of data. As a result, the electric energy consumption and throughput of each server is improved in the EEPR scheduler than the EEPO scheduler. In the evaluation, we show the lower electric energy is consumed by servers and each transaction takes the shorter execution time in the EEPR scheduler than the EEPO scheduler.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Energy-Aware Concurrency Control with the Purpose and Read-Data-From Serializability

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tomoya Enokido,
  • Dilawaer Duolikun,
  • Makoto Takizawa

摘要

In this paper, we newly propose the Purpose and Read-data-from (RDF) (RP-) serializability to provide energy-efficient application services. In the PR-serializability, transactions are serialized with the significant order of purposes and the RDF relation of transactions is maintained. The Energy Efficient Purpose and RDF (EEPR) scheduler is proposed to realize the PR-serializability by improving the Energy Efficient Purpose Ordering (EEPO) scheduler proposed in our previous studies. In the EEPR scheduler, each read method can be performed on each object without blocking by any other method reading the latest committed version of data. As a result, the electric energy consumption and throughput of each server is improved in the EEPR scheduler than the EEPO scheduler. In the evaluation, we show the lower electric energy is consumed by servers and each transaction takes the shorter execution time in the EEPR scheduler than the EEPO scheduler.