cclientserver

8/11/2019 cclientserver

1/9

Clientcentric consistency

models


2/9

Distributed Systems 2

Client-Centric Consistency models

Many systems: one or few processes perform updates How frequently should these updates be made available to other

read-only processes?

Examples:

DNS: single naming authority per domain Only naming authority allowed updates (no write-write conflicts)

How should read-write conflicts (consistency) be addressed?

NIS: user information database in Unix systems

Only sys-admins update database, users only read data

Only user updates are changes to password


3/9


Eventual Consistency

In absence of updates, all replicas converge towards identical copies Only requirement: an update should eventually propagate to all replicas Cheap to implement: no or infrequent write-write conflicts

Things work fine so long as user accesses same replica

What if they dont:


4/9


Semantics of Client-Centric Models

Assume read operations by a single process Pat two differentlocalcopies of the same data store Four different consistency semantics

Monotonic reads

Once read, subsequent reads on that data items return same or morerecent values

Monotonic w r i tes

A write must be propagated to all replicas before a successive write by thesame process

Resembles FIFO consistency (writes from same process are processed insame order)

Read you r wri tes: read(x) always returns write(x) by that process

Writes fo l low reads: write(x) following read(x) will take place on sameor more recent version of x


5/9


Monotonic Reads

The read operations performed by a single process Pat two differentlocal copies (L1 and L2) of the same data store.

Where xi denotes the version of x at local copy Li, and WS representsa write sequence, WS(x1; x2) denotes that x1 version is formed beforex2.

Ex: a user reads email x1 in New York, and then flies to Toronto, openthe copy of email box there, monotonic reads consistency guaranteesthat x1 will be in the mail box in Toronto.

A monotonic-read consistent data store A data-store that is not monotonic-read


6/9


Monotonic Writes

The write operations performed by a singleprocess Pat two different local copies of the samedata store

Resembles to PRAM, but here we are consideringconsistency only for a single process (client)instead of for a collection of concurrent processes.

A monotonic-write consistent data store A data store that is not monotonic-write


7/9


Read-Your-Writes

Closed related to monotonic reads

A write operation is always completed before asuccessive read operation by the same process

Ex: editor and browser, if not integrated, you maynot read-your-writes of an HTML page

Read-your-writes consistency Non Read-your-writes consistency


8/9


Write-follow-reads

Writes-follow-reads consistent Non Writes-follow-reads consistent

Updates are propagated as the result of previous read operation

Any successive write operation on x by a process will be performed on

a copy of x that is most recently read by that processEx: comments on news group, let A an article read recently, R the

response to that article, then R must follows A.


9/9

Thank You

Documents

cclientserver