 |
 |
 |
 |
||
 |
||
 |
||
 |
|
|
|
|
||
|
||
|
||
|
The design goal for voice networks is traditionally 99.999% or five 9s of availability. Though the Internet Protocol was designed to provide best effort service, IP networks are now an essential part of our communications infrastructure, with corresponding expectations of high availability. Yet, in many ways, management of IP networks remains a challenge, based more on ad-hoc tools than well-established principles. IP networks are prone to failures not only as a result of hardware faults, but also due to software bugs, misconfiguration, human errors, malicious users, and inadequate network management tools, exacerbated by the huge scale and diversity of the networks. In addition, with the growth of IP overlay applications, service providers must take into account both the health of the network and that of the (typically) distributed applications infrastructure, raising the ante from 'simple' network management to end-to-end service management. As a result, while IP networks and applications generally work well, the 'last 9' is an elusive goal and the challenges continue to increase. This workshop broadens the theme of INM'06 to explore both network and application service management, in the context of user expectations of high availability. The scope ranges from large scale backbone networks to end-to-end application performance and availability.
We are interested in questions such as the following: What metrics should be used to measure network and service availability and performance? How can overlay applications be designed to be resilient to failures, or to be managed effectively in the presence of failures? Given that routers and servers fail and are taken out of service periodically, what techniques for planned maintenance, redundancy and fail-over can be used, and what is their performance? What approaches are needed for network and software configuration given the thousands of elements and software modules that must be patched and updated in a live network? Given the large volumes of 'telemetry' data that is available from networks and servers, what tools can be used to facilitate rapid fault detection, isolation and remediation?
Areas of interest include, but are not limited to:
The workshop solicits original papers on completed work, position papers, and/or work-in-progress papers on the challenges raised above. Papers that bring out new and interesting approaches at an early stage of their development are very welcome.
Submissions must be no greater than 6 pages in length, must be a PDF
file and must follow the formatting
guidelines. Submissions that deviate from these guidelines will
be rejected without consideration. Note that those guidelines state
that submissions should
not be
formatted using the SIG style sheets: instead use LaTeX article
style or equivalent to produce 10pt double-column output with numbered
pages. Style sheets and other aids for
authors producing camera ready copy may be found here.
Reviews will be single-blind: author names and affiliations should be
included in the submission. Papers may be submitted at http://edas.info/ as follows:
1. Create a personal account on EDAS if you do not already have one, and log in.
2. Click on Submit paper tab and choose INM'07 from the list.
3. Register the paper with a short abstract.
4. Upload the PDF of the paper.
Authors of accepted papers are expected to present their papers at the workshop. Submissions must be original work not under review for any other workshop, conference, or journal.
| Abstract Submission: | Thursday, April 26, 2007 |
|---|---|
| Paper Submission: | (extended to) Monday, May 7, 2007 midnight Eastern Time USA |
| Acceptance Notification: | Thursday, May 31, 2007 |
| Camera Ready Due: | Wednesday, June 13, 2007 |
| Workshop: | Friday, August 31, 2007 |
 |
|
Home | Call For Papers | Committee | Programme | SIGCOMM 2007 | INM 2006 |
