On multi-version coding for distributed storage

Zhiying Wang, Viveck Ramesh Cadambe

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

The multi-version coding problem described previously by Wang and Cadambe is motivated by applications to distributed storage and computing. Here, we consider a modification to the previously described multi-version coding problem that retains the essence of the earlier definition, and show that our modification leads to a reduced storage cost. We consider a setting where there are n servers that aim to store ν versions of a message, where there is a total ordering on the versions from the earliest to the latest. We assume that each message version has size log2 M bits. Each server can receive any subset of the ν versions and stores over an alphabet of size q a function of the message versions it receives. The (n, c, ν, M, q) multi-version code we consider ensures that, a decoder that connects to any c of the n servers can recover the message corresponding to the latest common version stored among those servers, or a message corresponding to a version that is later than the latest common version. Unlike our earlier paper, we allow for the message version that is decoded to be one that is later than the latest common version. Through an achievable scheme and a tight converse, we describe the optimal multi-version code for ν = 2 versions from the perspective of the storage cost log2 q/ log2 M. In particular, we show that for ν = 2, the optimal multi-version code has a storage cost of 2/c+1 when c is odd and 2(c+1)/c(c+2) when c is even. We also present achievable code constructions for arbitrary values of the parameter ν.

Original languageEnglish (US)
Title of host publication2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages569-575
Number of pages7
ISBN (Electronic)9781479980093
DOIs
StatePublished - Jan 30 2014
Event2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014 - Monticello, United States
Duration: Sep 30 2014Oct 3 2014

Publication series

Name2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014

Other

Other2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014
CountryUnited States
CityMonticello
Period9/30/1410/3/14

Fingerprint

Servers
Costs

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Computer Science Applications

Cite this

Wang, Z., & Cadambe, V. R. (2014). On multi-version coding for distributed storage. In 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014 (pp. 569-575). [7028506] (2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ALLERTON.2014.7028506
Wang, Zhiying ; Cadambe, Viveck Ramesh. / On multi-version coding for distributed storage. 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014. Institute of Electrical and Electronics Engineers Inc., 2014. pp. 569-575 (2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014).
@inproceedings{95900708f284434c82f286dd69e82f11,
title = "On multi-version coding for distributed storage",
abstract = "The multi-version coding problem described previously by Wang and Cadambe is motivated by applications to distributed storage and computing. Here, we consider a modification to the previously described multi-version coding problem that retains the essence of the earlier definition, and show that our modification leads to a reduced storage cost. We consider a setting where there are n servers that aim to store ν versions of a message, where there is a total ordering on the versions from the earliest to the latest. We assume that each message version has size log2 M bits. Each server can receive any subset of the ν versions and stores over an alphabet of size q a function of the message versions it receives. The (n, c, ν, M, q) multi-version code we consider ensures that, a decoder that connects to any c of the n servers can recover the message corresponding to the latest common version stored among those servers, or a message corresponding to a version that is later than the latest common version. Unlike our earlier paper, we allow for the message version that is decoded to be one that is later than the latest common version. Through an achievable scheme and a tight converse, we describe the optimal multi-version code for ν = 2 versions from the perspective of the storage cost log2 q/ log2 M. In particular, we show that for ν = 2, the optimal multi-version code has a storage cost of 2/c+1 when c is odd and 2(c+1)/c(c+2) when c is even. We also present achievable code constructions for arbitrary values of the parameter ν.",
author = "Zhiying Wang and Cadambe, {Viveck Ramesh}",
year = "2014",
month = "1",
day = "30",
doi = "10.1109/ALLERTON.2014.7028506",
language = "English (US)",
series = "2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "569--575",
booktitle = "2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014",
address = "United States",

}

Wang, Z & Cadambe, VR 2014, On multi-version coding for distributed storage. in 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014., 7028506, 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014, Institute of Electrical and Electronics Engineers Inc., pp. 569-575, 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014, Monticello, United States, 9/30/14. https://doi.org/10.1109/ALLERTON.2014.7028506

On multi-version coding for distributed storage. / Wang, Zhiying; Cadambe, Viveck Ramesh.

2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014. Institute of Electrical and Electronics Engineers Inc., 2014. p. 569-575 7028506 (2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - On multi-version coding for distributed storage

AU - Wang, Zhiying

AU - Cadambe, Viveck Ramesh

PY - 2014/1/30

Y1 - 2014/1/30

N2 - The multi-version coding problem described previously by Wang and Cadambe is motivated by applications to distributed storage and computing. Here, we consider a modification to the previously described multi-version coding problem that retains the essence of the earlier definition, and show that our modification leads to a reduced storage cost. We consider a setting where there are n servers that aim to store ν versions of a message, where there is a total ordering on the versions from the earliest to the latest. We assume that each message version has size log2 M bits. Each server can receive any subset of the ν versions and stores over an alphabet of size q a function of the message versions it receives. The (n, c, ν, M, q) multi-version code we consider ensures that, a decoder that connects to any c of the n servers can recover the message corresponding to the latest common version stored among those servers, or a message corresponding to a version that is later than the latest common version. Unlike our earlier paper, we allow for the message version that is decoded to be one that is later than the latest common version. Through an achievable scheme and a tight converse, we describe the optimal multi-version code for ν = 2 versions from the perspective of the storage cost log2 q/ log2 M. In particular, we show that for ν = 2, the optimal multi-version code has a storage cost of 2/c+1 when c is odd and 2(c+1)/c(c+2) when c is even. We also present achievable code constructions for arbitrary values of the parameter ν.

AB - The multi-version coding problem described previously by Wang and Cadambe is motivated by applications to distributed storage and computing. Here, we consider a modification to the previously described multi-version coding problem that retains the essence of the earlier definition, and show that our modification leads to a reduced storage cost. We consider a setting where there are n servers that aim to store ν versions of a message, where there is a total ordering on the versions from the earliest to the latest. We assume that each message version has size log2 M bits. Each server can receive any subset of the ν versions and stores over an alphabet of size q a function of the message versions it receives. The (n, c, ν, M, q) multi-version code we consider ensures that, a decoder that connects to any c of the n servers can recover the message corresponding to the latest common version stored among those servers, or a message corresponding to a version that is later than the latest common version. Unlike our earlier paper, we allow for the message version that is decoded to be one that is later than the latest common version. Through an achievable scheme and a tight converse, we describe the optimal multi-version code for ν = 2 versions from the perspective of the storage cost log2 q/ log2 M. In particular, we show that for ν = 2, the optimal multi-version code has a storage cost of 2/c+1 when c is odd and 2(c+1)/c(c+2) when c is even. We also present achievable code constructions for arbitrary values of the parameter ν.

UR - http://www.scopus.com/inward/record.url?scp=84946687680&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84946687680&partnerID=8YFLogxK

U2 - 10.1109/ALLERTON.2014.7028506

DO - 10.1109/ALLERTON.2014.7028506

M3 - Conference contribution

T3 - 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014

SP - 569

EP - 575

BT - 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Wang Z, Cadambe VR. On multi-version coding for distributed storage. In 2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014. Institute of Electrical and Electronics Engineers Inc. 2014. p. 569-575. 7028506. (2014 52nd Annual Allerton Conference on Communication, Control, and Computing, Allerton 2014). https://doi.org/10.1109/ALLERTON.2014.7028506