|
Minutes From The September 8, 2005 Bimonthly meeting
|
|
Agenda
| Participants
|
- Agenda bashing
- EDDY early adopter target Identification
- kicking the tires and report on basic functionality
- expanding the development to other organizations
- Diagnostic data life cycle
- EDDY Progress to Date
|
- Chas DiFatta - CMU (chair)
- Mark Poepping - CMU
- Steve Olshansky - Internet2
- George Brett - Internet2 (scribe)
|
1. Agenda bashing
2. EDDY early adopter target Identification
a. kicking the tires and report on basic functionality
b. expanding the development to other organizations
Chas talked about updating the mailing list and checking to see who's on it.
He then talked about meeting with Paul and having a good conversation. He said Paul had some good ideas and areas he wanted to dig into. He said Michael Gettes will be leading the early adopter activities at Duke Unv.
3. diagnostic data life cycle
Chas said that a number of people were excited with the idea of a diagnostic data life cycle. CMU is among those interested. The question now is now the information will be cleaned and what would be retained. EDDY will help increase interest in these issues.
Chas - number people talked with Chas - excited about diagnostic data life cycle. Ken CMU others. How should be scoured, what should hang around... I think EDDY will raise volume on these issues.
4. EDDY progress to date
Chas said that current work is focused on normalization. At the moment EDDY can handle 6200 flows per second. He said that this is for demonstration purposes right now and just using MySQL rather than a high performance storage agent. The user interface will be a web interface to get this into the framework quickly so people can comment on it.
Chas pointed out that unlike earlier with display just on a laptop now 80% of the code will be re-usable. The focus is to have something in place for the Internet2 Fall Member Meeting.
The developers are working on a forensic query builder and a query navigator, that will use a browser to drill down thru Gigabytes of data which will include
an event viewer to look at a very granular level. The Flow data will include - source destination, port numbers, packet count, and protocol -- one can click on each and get much more detail about flow.
Other thing on development is the packaging. They are working on a management agent to manage agents on Unix host. Chas said they had a good work study student who helped put this into good JAVA framework which will permit quick builds later on.
Later they will be working on high end storage agent and documentation.
Steve asked how baked is it at the moment? Chas responded that it is baked, visually appealing, and has good work flow, but that stated the developers were concentrating on the demos for the fall member meeting. Much had to be done to get the release into a form that was developer(Java)/adopter friendly with respect to documentation and extensibility. I.e. polishing the Java framework. Chas said that the challenges with the diagnostic applications for the demo is not just the user interface but also the issue of managing large data sets. Mark pointed out that it is baked as a proof of concept tool for now. It's not an end product for developers. The next stage is to produce something that early adopters can use and comment on. We'll get started on this after the fall member meeting
The call ended.
|