I'm excited about this workshop, which is June 17, 2013 and co-located with HPDC. A couple of us from CTSC, myself and Jim Marsteller, are on the program committee. Please see the workshop website for more information and note that abstracts are due February 4th, 2013.
From Scott Campbell and Aashish Sharma's description:
Providing effective and non-intrusive security within a HPC
environment provides a number of challenges for both researchers and
operational personnel. What constitutes HPC has expanded to include
cloud computing, 100G networking, cross-site integration, and web 2.0
based interfaces for job submission and reporting, increasing the
complexity of the aggregate system dramatically. This growing
complexity and it's new issues is set against a backdrop of routine
user and application attacks, which remain surprisingly effective over
time.
The CLHS workshop will focus on the problems inherent in securing
contemporary large-scale compute and storage systems. To provide some
clarification we have broken this out into four general areas or
questions. First is Attribution: who is doing what in terms of
process activity and/or network traffic? Second is looking beyond the
interactive nodes: what is going on in the computing pool? Third
involves job scheduler activity and usage: what is being run, how has
it is been submitted and is this activity abnormal? Finally a more
philosophical topic of why securing complex systems is so difficult
and what can be done about it. While these specific areas are
interesting starting points for papers and presentations, any original
and interesting topic will be considered.
This year there will be two separate tracks for paper submission:
* Research Paper track
* State of the Practice
We will ask participants in the research paper track to add a section
describing in some detail their ongoing and future data needs. This
is principally to help researchers articulate the details of their
data oriented needs, as well as improving communications between data
generators and consumers.
For the State of of the Practice papers, the focus will be on the
resolution of specific issues - ideally those identified in the
Overview section, but really any significant problem which is endemic
to the HPC domain. Within the paper an explanation and exploration of
the issue, resolution description and a numerical analysis showing
that the proposed issue resolution was successful. Like in the
Research Papers track, it would be desirable to add a section on data
resources that might be available to researchers - either individually
or in larger sets.
