CEC 2007 Breakout Sessions in a Nutshell
By rajeshr on Oct 09, 2007
Session 1: DTrace & Java: Spanning the Observability Gap
Speakers: John Haslam & Simon Ritter
The one-liner summary of this session would say, "Using DTrace to look at Java Applications." It appears that the speakers were misinformed about the venue and they had to run around figure out their actual venue for the talk. There was a slight delay in starting up the program. I wanted to know more on DTrace and that is why I chose to sit through this session. Well, this session wasn't fully about the DTrace tool, but about leveraging the power of DTrace to observe/debug a Java Application. Do I didn't fully get what I actually wanted. But to say the least, I got some of my DTrace fundamentals straight after attending this session. The presenters repeatedly pointed out that the tools like Netbeans do have a powerful profiling mechanisms to count the number of methods and objects in an application, but was limited to that and could not cut across the entire software stack and reach the Kernel to figure out what Kernel is doing when an application is running. They demonstrated this feature by running a game and occasionally showing us the piece of code written in D script to scrutinize this application.
Another interesting aspect of this session was the introduction to DTrace grapher. I had not seen something of that sort before. The presenters pulled up a program to create rectangular boxes, and each attempt to create a rectangular box, resulted in a massive amount of calls in the Kernel. The same was displayed in something like a state transition diagram, which looked quite scary. Simon showed us a printed copy of one such diagram, which spanned a chart, bigger than a architectural plan for a star rated hotel (a little bit of exaggeration, I hope is forgiven)!!
Session 2: How to take advantage of the Dynamic Reconfiguration of the Sun SPARC Enterprise Mx000 (OPL) Servers
Speakers: Stephane DUTILLEUL & Ramon Garcia
This was of course a session that wasn't in my agenda. I got inside the venue for this session, only because I had some time at hand before I attended my next session. I didn't have much background on the High End Servers and for that reason, it was difficult for me to pay undivided attention to what the speakers had to say. From whatever I understood, the following is what I inferred: Dynamic Reconfiguration is not so pretty in Mx000 in this point of time, but it is likely to get better in the next few weeks. I know this information isn't going to help any, but I am afraid that's all I can provide from my side.
Session 3: Directory Server 6- Tips, Tricks and Secrets
Speakers: Ludovic Poitou & Terry Sigle
This was a session that I was looking forward to attend and I must say that I was very satisfied with the contents of this session. The presenters dealt the session with extreme professionalism and was very careful in discussing the basics about the Directory Server before getting into its internal. And if you want to verify my attendance for this session, you could check out the blogs of one of the presenters here and you would see me in the photograph
A few key points that I would wish to share here is that the DB technology underneath the Directory Server is sleepy cat (precisely the version 4.4), which is open source (Berkeley DB) database, now acquired by Oracle(an attempt to access the site sleepycat.com would redirect one to the Oracle website). It, however, continues to be OpenSource.The speakers referred about a tool db_stat which is a sleepy cat utility, but unbundled with the Directory Server EE 6.0. If required, this could be downloaded and used. A word of caution: the version of this has to be same as that of the DS. I recall the command syntax as:
db_stat -N -d filename
Some other recommendation that the presenters mentioned in their presentation include:
\* When you design the schema, make the attribute name short \* Spaces in the DN (after each comma) are not recommended though it an accepted syntax \*Avoid duplications with different representation
\*XML and non-ASCII values are base 64 encoded (+33% in size)
I felt the executive summary of the session on Directory Server 6.0 this afternoon was: For a successful deployment of Directory Server, understanding its internals is a must. And this session certainly provided some valuable pointers to achieve that objective. Should you feel interested in attening an SLS course on Directory Server, please go through the course description and contents here
Session 4: Project Indiana
Presenter: Glynn Foster
This was a very cool session. Firstly, Glynn accepted the fact that there are a lot of people out there who are struggling with the installation of Solaris 10. Then he went on and talked about the history of OpenSolaris a bit after which he discussed about tomorrows opportunity. Three key points from tomorrows opportunity includes the following:
\* Ease of Access
\* Developer Familiarity
\*User Community - Approachability for new users
Glynn summarized the goals of Project Indiana (re-distributable version on Open Solaris expected in the future):
\* An OpenSolaris binary distribution
\*Focuses on Solaris unique features
\*Easy to acquire
\*More familiar and discoverable
\*Built by community
\*Single CD install image
\* Core OS
>Kernel, System Libraries, System utilities
\* Network based package management system
\* ZFS as default filesystem
Glynn kindly demonstrated to us a quick installation of Solaris Express DE using the new installer (boy, that was super fast and invoked ovation) and also the package management system.
All the while Glynn addressed the audience for OpenSolaris as Developers, which prompted on of the attendees to ask this question: "You have been repeatedly mentioning the audience of OpenSolaris as Developers. What about the 14 year old - 15 year old folks, who are the future developers? When would they be able to use OpenSolaris?" Glynn just smiled and said, "Well, I don't know." Neither do I. So let's leave that topic there and move on.
Session 5: Solaris 10 Security Features/Futures
Speakers: Valerie Fenwick & Christoph Schuba
I must confess that I didn't have the sufficient background to sit through this session and hence the take back from this session for me was minimal. But both of the speakers kept the event lively and talk eloquently about the improvements in the Cryptography Framework and also the plans for Virtualization in Solaris (Christoph took up the second part). One of the important points that I could pick up from Valerie's presentation was the major developments in the field of Kerberos including the ease of set up. I wish I could read a bit more about the Cryptography framework, then I wouldn't been able to follow lot many things in this session. So here is an advise to myself: Have sufficient background information before listening to the authorities of specific subjects. Oh by the way, SLS offers you an exam on Solaris 10 Security Administrator, the objectives of which you may find here. Should you feel interested in taking up an SLS course on Solaris Security Administrator, have a look at the course description on this website
I am looking forward to the other exciting sessions tomorrow.