Distributed consensus

PDF Publication Title:

Distributed consensus ( distributed-consensus )

Previous Page View | Next Page View | Return to Search List

Text from PDF Page: 020

20 1.6. SCOPE & LIMITATIONS Reconfiguration – We assume a fixed and known set of participants each with a unique identifier. Reconfiguration is discussed extensively in the literature and is a compo- nent of many algorithms. Examples include Stoppable Paxos [MLZ08], VRR [LC12, §7], Raft[OO14, §6]. Weakened semantics – We do not support operations with weakened semantics such as stale reads or operations which rely on synchrony or bounded clock drift for safety such as master leases [Bur06, VRA15]. Implementation details – We assume unbounded storage, representation of arbitrary values, no corruption to state or messages. Participants may stop and restart. Upon restarting, the persistent state is unchanged, non-persistent state is re-initialised and the algorithm is executed again from the beginning. The pseudocode provided in this thesis is assumed to be executed in order by a single thread and each line is executed atomically. Writes to state must be completed before proceeding, including writes to persistent storage. This can be achieved by techniques such as Write-ahead logging [MHL+92]. Reads from state must always return an up to date value. Partial ordering – Our algorithms decide a single value (or decide a totally ordered, infinite sequence of values). We do not consider agreement over multiple series of values, partially ordered sequences [Lam05b] or finite sequences [MLZ08]. Progress in practice – Participants may operate at arbitrary speeds. Messages are eventually delivered but there is no bound on the time for the communication channel to deliver messages. Messages may be delivered out-of-order or multiple times. However, the progress of the algorithms depends on extensive assumptions, including synchrony and timing. We prove the progress for our algorithms under these assumptions however they are not minimal. Specific systems – All algorithms are provided as high-level representations, not concrete protocols or implementations. To remain applicable to a range of existing and further systems, we do not optimise for particular systems or workloads as has been the subject of extensive research. For example, Ring Paxos[MPSP10] and Multi-Ring Paxos[MPP12] optimise for networks providing IP multicast.

PDF Image | Distributed consensus

PDF Search Title:

Distributed consensus

Original File Name Searched:

UCAM-CL-TR-935.pdf

DIY PDF Search: Google It | Yahoo | Bing

Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info

Cruising Review Topics and Articles More Info

Software based on Filemaker for the travel industry More Info

The Burgenstock Resort: Reviews on CruisingReview website... More Info

Resort Reviews: World Class resorts... More Info

The Riffelalp Resort: Reviews on CruisingReview website... More Info

CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com (Standard Web Page)