Printable Version

Forum for Proposals to iCSC2010

Name_of_posting_person: Francois Fluckiger
CSC_year: NA
Type_of_posting: Welcome message
Date: September 20, 2009
Time: 16:59

Welcome to CSC2008 and CSC2009 participants. I hope you enjoyed our school and you will be willing to suggest topics through this forum. However, I know that the current year is extremely busy for several of you (LHC start). Therefore, consider this as an opportunity but not as having any kind of moral obligation.
François, School Director

Name_of_posting_person: Maksym Borodin
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 24 September 2009
Time: 00:37

Message

I can propose a topic on improvement on performance of ROOT reading files from mass storage.

I've made some researches with dCache and found out that different combinations of basket size (used, when branch was created) and dCache specific settings for readahed can give good change in timing, when ROOT ntupples are read from dCache.

Now I'm planning to extend this research and find out, is there are dependencies on size of object, stored in branch. If it would be useful, I can give a presentation of my results with some obtained numbers.

Name_of_posting_person: Tim München
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 24 September 2009
Time: 10:05

Message

My proposal: "Living OO Design: Design Patterns and best practices"

I held a lecture incl. exercises and an exam for master-students on that topic on our University of Applied Sciences, so there is quite a bit of material prepared already; this would've to be compressed a bit to fit into 2 (or so) lectures.

The material covers the following topics:

• A crash course of UML for developers

• OO best practices (dubbed "The 10 commandments of OO design")

• Selected design patterns from the GoF-catalogue

The target audience would be students and physicists with basic OO-programming knowledge. The examples are implemented in Java.

The goal of the lecture is to make people aware of common OO-design pitfalls and how to fully utilize existing design patterns instead of reinventing the wheel.

Name_of_posting_person: James Thorne
CSC_year: CSC2008
Type_of_posting: I have a specific topic to propose
Date: 01 October 2009
Time: 11:04

Message

Data storage is an important part of the Grid. Tuning the operating system, file system and RAID controller and other parameters to achieve the the optimum balance between performance, reliability and data integrity is essential to create a responsive disk system for Grid data storage.

Having worked for two years as an administrator at an LCG Tier1 with specific responsibility for storage, I propose to talk about ways to improve storage system performance, reliability and data integrity through tuning, monitoring, testing and other tools, including:

- RAID (and typical controller options)

- File systems

- Linux operating system tunables

- Monitoring storage systems

- Testing storage systems

- Tools and "early warning" systems

Name_of_posting_person: David Horat
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 06 October 2009
Time: 11:36

Message

Following the presentation I made at the CSC 2009, "Response Time Optimization in Web Applications", and considering people's interest in the topic, I would like to further analyze the tools and techniques that we can use to optimize web applications.

Thus, I propose a 2-hour lecture with the following content:

- Introduction

- Network technologies

- Application architecture

- Javascript and CSS

- Other components - Tools review

- Real world example

- Conclusions

- Q&A Note:

This summary is just a proposition, open to suggestions, which may be altered in the final presentation. Regards, David

Name_of_posting_person: David Svantesson
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 13 October 2009
Time: 09:37

Message

Seeing that CVS was presented at iCSC2005, one idea could be an (~1 hour) presentation on the newer git (http://git-scm.com/ if it could be fit into a larger theme.

Distributed revision control systems (and git in particular) are rather different from traditional ones like cvs and svn and brings (potentially) more freedom to the work pattern. It fit equally well for personal use as well as in large collaborations.

Example of topics that could be covered:

• Distributed revision control systems, why we need it!
• Basic usage and some more advanced, including
  • solving merge conflicts
  • changing commit history

• Examples of good (/not so good?) workflows using git (especially for collaboration using git)
  • How can I benefit using git even if my group uses cvs/svn
  • People don't want to learn new change to git and let them continue with cvs/svn
  • Why/when not to use git (drawbacks).

The target audience would be people with some (little) previous knowledge of revision control systems.

Name_of_posting_person: Francois Fluckiger
CSC_year: NA
Type_of_posting: Comments from Director after 1st round of postings
Date: October 13, 2009
Time: 11:15

Many thanks to the 5 people who already posted. A few comments at this point (for past and future postings):

Remember: iCSC is not a mini -conference. As CSCs, it is an academic event. Therefore:

• The topic should be novel. Its context should be briefly presented (state of the art; how what I teach fits in it, ...)
• The lectures should not focus on what you do, on your project, defending a thesis, presenting a system you have developed .
• Of course, if you are knowledgeable on the academic topic that you propose to teach, this is probably because you are working practically on it. But if you refer to your work, this should be as an illustration of the academic topic, not as the main substance of the class

From a practical angle

• Lectures will last 55 min.
• There will be no mandatory registration (except to get the booklet).
• The programme will be a menu: Participants may attend a single lecture.
• As  a consequence:
  • Each lecture should be reasonably self-contained (e.g. starting with a very short summary of the previous lecture, if this part of a series)
  • When we publish the final programme, the targeted audience, the pre-requisites and the level of each individual lecture should be clearly indicated  (not necessarily in the proposal)
    (e.g."This lecture is an introduction to the topic, and indented to those who are less familiar with ..." or "This lecture goes more in depth on ...  and assumes good familiarity with ...").
• We cannot organize exercises.

Finally, proposal should as far as possible indicate the proposed (approximate) number of lectures.

Thanks again for your interest in iCSC. François, School Director

Name_of_posting_person: Svetozar Kapusta
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 29 October 2009
Time: 18:12

Message

I would like propose to organize one day for the Data Management and Data Bases Theme consisting of few introductory and advanced database-related lectures. It could be similar to what was already organized for the iCSC2005.

Please let me know if someone is willing to join this effort and give a lecture on his preferred database-related topic.

Suggestions and comments are more than welcome.

Thanks! Sveto

Name_of_posting_person: Uwe Westerhoff
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 29 October 2009
Time: 23:54

Message

I propose a one hour lecture about the Simple Network Management Protocol (SNMP) and its usage to monitor and operate systems. The SNMP defines standards to query information from the hardware and to send commands to the hardware. It is supported by almost all hardware attached to a network, from the WLAN access point you have at your home to big RAID systems or routers. Even very specialized hardware like the low voltage power supplies used in the ALICE experiment can be operated via SNMP. A popular software which uses SNMP (among other protocols) to monitor hardware is Nagios.

Therefore I propose a lecture which will cover the following topics:

• SNMP: general design
• Assets and drawbacks of the different SNMP versions
• Management Information Bases (MIBs) (important to understand how information is stored and queried via SNMP)
• The ALICE low voltage power supplies as an example for using SNMP
• Introduction to Nagios
• How to use Nagios to monitor hardware via SNMP

Name_of_posting_person: Malte Nuhn, Luis Fernando Muñoz Mejías
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 30 October 2009
Time: 00:29

Message

We want to present tools and techniques to extract and gather information from a computer system. We'll focus on two use cases:

• When developing software:
  • log4xxx + syslog (or why you should stick to standards and not invent your own log system)
• When running binaries you cannot change:
  • Systemtap + Kernel

The information collected can be then used for debugging, security purposes or performance analysis. We'll use real examples to illustrate these tools.

We will explicitly skip profilers, as they were already covered by the standard CSC.

Name_of_posting_person: Malte Nuhn, Luis Fernando Muñoz Mejías
CSC_year: CSC2009
Type_of_posting: I am commenting on a previous posting
Date: 30 October 2009
Time: 11:40

Message

Forgot to mention, we plan three 55-minute sessions. Logging in the source code and syslog (filters, security...) may need one session, low-level details such as the kernel and Systemtap need two sessions.

Name_of_posting_person: Benjamin Radburn-Smith
CSC_year: CSC2009
Type_of_posting: I have a specific topic to propose
Date: 31 October 2009
Time: 00:04

Message

Hello all,
 I could give a single lecture exploring some of the recently developed visualisation techniques such as parallel coordinates and the grand tour. Parallel coordinates was briefly mentioned at the CSC2009 school - however I would like to explain in more detail how the plot could be used and its advantages over conventional techniques.

I would also like to describe and demonstrate the grand tour technique and how it could be combined with the parallel coordinate plot allowing the user to find interesting characteristics of the data.

I should point out that even though that some of these topics were mentioned at the CSC2009 - this is what my project focuses on.

Thanks. Benjamin

Name_of_posting_person: Francois Fluckiger
CSC_year: NA
Type_of_posting: Welcome message
Date: October 31, 2009
Time: 13:30

Many thanks to all those who posted proposals.
Initial deadline was today, but we wish to extend it by another 3 days (i.e. last day to post is now Tuesday 3 November ). This may also give time to those interested to comment on Sveto's call for partners above.
François, School Director

Name_of_posting_person: David Svantesson
CSC_year: CSC2009
Type_of_posting: I am commenting on a previous posting
Date: 02 November 2009
Time: 20:40

Message

There are some suggestions related to development and improvement of software. Maybe we can get together and make a theme of it?

The suggestions (so far) I think could fit into this theme:

• Tim: Living OO Design
• David H: Optimisation of web applications
• David S: Distributed revision control with git
• Malte&Luis: software logging, system logs

Total time indicated by suggesters is 8h. Although the topics might seem a bit diverse, they have in my opinion become increasingly important in recent years or present novel technologies.