Ian F. Adams
I'm Ian Adams, a fifth year computer science grad student at University of California, Santa Cruz. I work with
Professor Ethan Miller in the Storage Systems Research Center (SSRC). My current research interests are in archival storage, and the compute-restore trade-off, see the paper titled "Maximizing Efficiency By Trading Storage for Computation" below for more details on the latter.
I may be contacted at iadams AT soe DOT ucsc DOT edu
On campus, I'm usually at the main SSRC lab in E2-381, feel free to stop by!
Peer Reviewed Publications
Evolutionary Trends in a Supercomputing Tertiary Storage Environment. Please contact me if you need an early copy
Joel C. Frank, Ethan L. Miller, Ian F. Adams, Daniel C. Rosenthal,, To appear in Proc. of MASCOTS 2012, August 2012.
Ian F. Adams, Ethan L. Miller, Mark W. Storer, "Analysis of Workload Behavior in Scientific and Historical Long-Term Data Repositories", ACM journal Transactions on Storage 8(2), May 2012.
Ian F. Adams, Ethan L. Miller, Mark W. Storer, Proceedings of The 18th Annual Meeting of the IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, August 2010.
Ian F. Adams, Darrell D. E. Long, Ethan L. Miller, Shankar Pasupathy, Mark W. Storer, "Maximizing Efficiency By Trading Storage for Computation", Proceedings of the Workshop on Hot Topics in Cloud Computing (HotCloud '09), June 2009.
Mark W. Storer, Kevin Greenan, Ian F. Adams, Ethan L. Miller, Darrell D. E. Long, Kaladhar Voruganti, "Logan: Automatic Management for Evolvable, Large-Scale, Archival Storage," Proceedings of the 2008 Petascale Data Storage Workshop (PDSW 08), November 2008.
Ian F. Adams, Ethan L. Miller, David S.H. Rosenthal, Technical Report UCSC-SSRC-11-07, May 2011.
Brian Madden, Ian F. Adams, Mark W. Storer, Ethan L. Miller, Darrell D. E. Long, Thomas Kroeger, Technical Report UCSC-SSRC-11-04, May 2011.
Ian F. Adams, Ethan L. Miller, Mark W. Storer, "Analysis of Workload Behavior in Scientific and Historical Long-Term Data Repositories", Technical Report UCSC-SSRC-11-01, March 2011
Andrew Leung, Ian F. Adams, Ethan L. Miller, "Magellan: A Searchable Metadata Architecture for Large-Scale File Systems," Technical Report UCSC-SSRC-09-07, November 2009.
These are posters that have been peer reviewed prior to presentation at a conference.
Class and Unpublished Work
These are class projects and other unpublished work from a variety of areas.
Automating Analysis of the Computation-Storage Tradeoff
This work was done as a class project for CMPS 232: Distributed Systems with
fellow SSRC student Brian Madden
Abstract:The standard paradigmin computing today is that of compute,
store, and recall. In this workflow the final result of a
computation is stored for later retrieval. Due to their massive
computational power, the rise of cloud and grid based
systems calls into question this compute, store, retrieve
paradigm. The computational power that can be allocated
on demand can allow datasets to be recomputed relatively
quickly, removing the need store rarely used intermediate
and final results.
There are several challenges that arise when considering
this store vs. recompute tradeoff however. First, it
must be determined if the trade-off is even feasible or desirable;
and second, what data-sets should be chosen for
storage to reduce costs and computation times. We focus
on the second of these challenges in this paper.
In our work here, we model large computations as annotated
directed acyclic graphs where by nodes represent
data, and edges represent processes. Using a series of simple
heuristic and randomized-search algorithms to select
a subset of nodes to store, we demonstrate it is possible to
yield significant savings in storage and computation costs
at the expense of increasing re-computation time.
Downlod pdf file of CMPS 232 term project writeup
2-Tier Leadership Election
This work was done as a class project for CMPS 221: Advanced Operating Systems
Abstract: Current leadership election algorithms do not explicitly plan for the failure of the leader, and thus must do system wide elections each time a leader fails. To this end we propose
an algorithm that removes the need for system wide elections after a leadership failure. Our technique has an initial system wide election to elect a leader, and after the leader is
confirmed chooses subordinates with which to form a clique. After the failure of the leader a subordinate assumes the leaders position, and a new subordinate is selected to take
the vacated place.We demonstrate the feasibility of this new scheme with a simple message passing simulation and found that our scheme cuts in half the number of messages required
to select and announce a new leader after a leadership failure.
Downlod pdf file of 2-Tier leadership Election
FOP and AOP:Benefits, Pitfalls and Potential for interaction
This work was a class project from CMPS 290G under Jim Whitehead and was done with a fellow grad student, Sigmon Myers.
We examined the FOP and AOP programming paradigms and created a new modeling notation for
combining them. Additionally we created a small example of a combined FOP and AOP program using
the AspectJ and FeatureIDE tools.
Download pdf file of FOP Project Writeup
The Quantum 8-Tile Puzzle: A Variation on the Classic 8 Tile puzzle.
This work was done as a term project for the CMPS 240-Artificial Intelligence course under Robert Levinson.
For this project I examined methods for solving a variation of the 8-Tile puzzle wherein each tile and the space can be broken
up into smaller constituent pieces and swapped in small increments.
Download pdf of CMPS 240 AI Project Writeup
Chance and Deception: Attacks That Rely On User Actions
This work was done as a class project for CMPS 223: Advanced Security
There are a variety of attacks that rely heavily on user
actions to succeed and dont require explicit installs or
on the part of the user. We examine 3 basic categories
of such attacks: Autorun attacks utilizing physical media
such as flash drives and CD-Roms, phishing attacks and
web based malware attacks that do drive-by-downloads.
Through surveying users we found the following: First,
many users are vulnerable to autorun based attacks, second,
substituting visually similar but semantically different
characters in hyperlinks provides a particularly effective
way of obfuscating a malicious websites hyperlink,
and third there appears to be an interesting logical disconnect
whereby web links advertised physically, but still of
unknown origin, are deemed more trustworthy than those
from an unknown email source.
We also attempted two simple proof of concept experiments
to demonstrate the feasibility of using physical advertising
through flyers and promotional CDs as attack
vectors to guide users to malicious websites. Both experiments
failed, with a minimal number of hits to the flyer
advertised website, and none using the CDs.
Download pdf of CMPS 223 term project write-up
And for the interested persons, a little more on me personally...