N. Hayashibara, "Accrual Failure Detectors," Ph.D. thesis, Japan Advanced Institute of Science and Technology (JAIST), 2004.
ID 58
分類 博士論文
タグ accrual failure detectors
表題 (title) Accrual Failure Detectors
表題 (英文)
著者名 (author) Naohiro Hayashibara
英文著者名 (author) Naohiro Hayashibara
キー (key) Naohiro Hayashibara
学校名 (school) Japan Advanced Institute of Science and Technology (JAIST)
出版社住所 (address)
刊行月 (month) 6
出版年 (year) 2004
URL https://dspace.jaist.ac.jp/dspace/bitstream/10119/956/3/2443paper.pdf
付加情報 (note) No. 133
注釈 (annote)
内容梗概 (abstract) Failure detection is a fundamental building block for ensuring fault tolerance in distributed systems.
Failure detector is a distributed entity and is consist of the set of failure detector modules.
Each module outputs the set of suspected processes. Failure detectors are basic components to
solve agreement problems (e.g., consensus, atomic broadcast, atomic commitment), especially
in asynchronous distributed systems. There are lots of approaches and implementations in failure
detectors. However, providing flexible failure detection in off-the-shelf distributed systems
is difficult.
Practical solutions to failure detection rely on some adaptive mechanism to cope with the
unpredictability of networking conditions. However, they lack the necessary flexibility to provide
failure detection as a system-wide service. In particular, traditional solutions take a “one
size fits all” approach, which prevents them from simultaneously supporting several distributed
applications with very diverse QoS requirements.
In this dissertation, we present a novel approach, called accrual failure detectors, that addresses
the flexibility issue. Accrual failure detectors are fundamental approaches for detecting
failures in point-to-point communication with a pair of nodes or processes. Conventional failure
detectors provide information of a boolean nature (i.e., suspect or not suspect). In contrast, accrual
failure detectors provide a value to express the confidence that a given process has crashed.
If the process actually crashes, the value accrues until eventually reaching any given threshold
set by applications, hence we named our failure detectors “Accrual failure detectors”. On the
other hand, each application has own threshold respect to its requirement. It suspects some
process itself according to the threshold and the value given by the failure detector module.
In the dissertation, we describe the concept and the definition of accrual failure detectors.
Then we present mechanisms and implementations of two instances of accrual failure detectors,
the ϕ-failure detector and the κ-failure detector. We also show our measurement of our failure
detectors and comparison of these failure detectors and other adaptive failure detectors.

論文電子ファイル hayashibara_phd_thesis.pdf (application/pdf) [一般閲覧可]
         title = {Accrual Failure Detectors},
        author = {Naohiro Hayashibara},
        school = {Japan Advanced Institute of Science and Technology (JAIST)},
         month = {6},
          year = {2004},
          note = {No. 133},