MPI-INF Logo
Publications

Server    domino.mpi-inf.mpg.de

Proceedings Article, Paper
@InProceedings
Beitrag in Tagungsband, Workshop

Author, Editor
Author(s):
Bringmann, Karl
Doerr, Benjamin
Neumann, Adrian
Sliacan, Jakub
dblp
dblp
dblp
dblp
Not MPG Author(s):
Sliacan, Jakub
Editor(s):
Fomin, Fedor V.
Freivalds, Rūsiņš
Kwiatkowska, Marta
Peleg, David
dblp
dblp
dblp
dblp
Not MPII Editor(s):
Fomin, Fedor V.
Freivalds, Rūsiņš
Kwiatkowska, Marta
Peleg, David
BibTeX cite key*:
BringmannDoerrNeumannSliacan2013
Title, Booktitle
Title*:
Online Checkpointing with Improved Worst-Case Guarantees
Booktitle*:
Automata, Languages, and Programming - 40th International Colloquium (ICALP 2013)
Event, URLs
Conference URL::
http://icalp2013.lu.lv/
Downloading URL:
http://arxiv.org/pdf/1302.4216v2
Event Address*:
Riga, Latvia
Language:
English
Event Date*
(no longer used):
Organization:
European Association for Theoretical Computer Science (EATCS)
Event Start Date:
8 July 2013
Event End Date:
12 July 2013
Publisher
Name*:
Springer
URL:
http://springer.com
Address*:
Heidelberg
Type:
Vol, No, Year, pp.
Series:
Lecture Notes in Computer Science
Volume:
7965
Number:
Month:
Pages:
255-266
Year*:
2013
VG Wort Pages:
27
ISBN/ISSN:
978-3-642-39205-4
Sequence Number:
DOI:
10.1007/978-3-642-39206-1
Note, Abstract, ©
(LaTeX) Abstract:
In the online checkpointing problem, the task is to continuously maintain a set of $k$ checkpoints that allow to rewind an ongoing computation faster than by a full restart. The only operation allowed is to replace an old checkpoint by the current state. Our aim are checkpoint placement strategies that minimize rewinding cost, i.e., such that at all times $T$ when requested to rewind to some time $t \le T$ the number of computation steps that need to be redone to get to $t$ from a checkpoint before $t$ is as small as possible. In particular, we want that the closest checkpoint earlier than $t$ is not further away from $t$ than $q_k$ times the ideal distance $T / (k+1)$, where $q_k$ is a small constant.

Improving over earlier work showing $1 + 1/k \le q_k \le 2$, we show that $q_k$ can be chosen asymptotically less than $2$. We present algorithms with asymptotic discrepancy $q_k \le 1.59 + o(1)$ valid for all $k$ and $q_k \le \ln(4) + o(1) \le 1.39 + o(1)$ valid for $k$ being a power of two. Experiments indicate the uniform bound $p_k \le 1.7$ for all $k$. For small $k$, we show how to use a linear programming approach to compute good checkpointing algorithms. This gives discrepancies of less than $1.55$ for all $k < 60$.

We prove the first lower bound that is asymptotically more than one, namely $q_k \ge 1.30 - o(1)$. We also show that optimal algorithms (yielding the infimum discrepancy) exist for all~$k$.
URL for the Abstract:
http://arxiv.org/abs/1302.4216
Download
Access Level:
Internal

Correlation
MPG Unit:
Max-Planck-Institut für Informatik
MPG Subunit:
Algorithms and Complexity Group
Audience:
experts only
Appearance:
MPII WWW Server, MPII FTP Server, MPG publications list, university publications list, working group publication list, Fachbeirat, VG Wort



BibTeX Entry:

@INPROCEEDINGS{BringmannDoerrNeumannSliacan2013,
AUTHOR = {Bringmann, Karl and Doerr, Benjamin and Neumann, Adrian and Sliacan, Jakub},
EDITOR = {Fomin, Fedor V. and Freivalds, Rūsiņš and Kwiatkowska, Marta and Peleg, David},
TITLE = {Online Checkpointing with Improved Worst-Case Guarantees},
BOOKTITLE = {Automata, Languages, and Programming - 40th International Colloquium (ICALP 2013)},
PUBLISHER = {Springer},
YEAR = {2013},
ORGANIZATION = {European Association for Theoretical Computer Science (EATCS)},
VOLUME = {7965},
PAGES = {255--266},
SERIES = {Lecture Notes in Computer Science},
ADDRESS = {Riga, Latvia},
ISBN = {978-3-642-39205-4},
DOI = {10.1007/978-3-642-39206-1},
}


Entry last modified by Adrian Neumann, 02/17/2014
Hide details for Edit History (please click the blue arrow to see the details)Edit History (please click the blue arrow to see the details)

Editor(s)
[Library]
Created
01/10/2014 11:16:28 AM
Revision
1.
0.


Editor
Adrian Neumann
Adrian Neumann


Edit Date
10/01/2014 13:25:14
10/01/2014 11:16:28