suncec2007

Okay, the flight back home starts to get a little bit annoying. I have to re-checkin at SFO again, while my baggage was checked in until Franfurt. So i have to checkin my baggage in Frankfurt to get it to Berlin. So i have to go to checkin at every airport of this journey.

It´s half five now ... my flight to SFO goes in 4 hours to board a 747 heading to Germarny there.. This was my trip to Las Vegas and to CEC2007. Was a really good event. I think i will write a summary on my flight back. Albeit it was nice here, it´s good to get at home now ... i want my own bed.

PS: I´ve made somewhat over 400 photos of the Bellagio water fountains in action yesterday evening. There some really good shoots. Will upload them back home.

How can you measure the interestingness of a keynote? It´s really simple. My colleague Gerd observed, that the more interesting a keynote was, the less notebooks were active. Thus the metric to measure the interestingness would be the count of notebook backlights. Or as I suggested to him for automating this metric: Simply count the active MAC addresses on the WLAN routers.

I was in the GSE All Hands Meeting. It´s a internal meeting, so i won´t write about it´s content. But one thing was really interesting: Most of the people looked so tired like i feeled yesterday. From this point it must been a really good party yesterday evening.

What´s the sense of having such a huge engineering event. It´s not a small or cheap challenge to get 4000 Engineers at one place. So ... what´s the benefit: Albeit most breakouts were really good, the keynotes were informative and inspiring , the trip to Las Vegas is a nice incentive.

But the core asset of this conference lies somewhere else It´s communication, the talks between the session in the floors. I had some really interesting discussions in informal talks at CEC. I have learned a thing or two but more important i´ve met people with different perspective , different challenges and different solutions on topics similar to those, which keeps me awake.

It was worth to travel 10 hours for this talks alone ... well ... to be exact .. it were 24 hours bed to bed ...

I´ve must been really tired yesterday. At least i woke up at 10 o´clock in my hotelroom yesterday evening sleeping left of the running notebook with a flashcard reader under my back and the pocketbook with my notes wrapped into the blanket ...

Five or six people already asked me, why i use my smartphone (an Nokia E61, when you look around at CEC it seems, that half the production of this phone has gone to Sun) in the sessions and type messages troughout the . It´s not that i´m ruining Sun by sending short messages to a love interest (okay, at least not at the keynotes, and i prefer IM for that ... national GSM short messages are expensive, the price for international ones is outright obscene ).

It´s much simpler ... i feed my blog with this phone. Well, not directly and not for the long articles, but when you look at the front page you see a block of five short messages on the top. This block is feeded by my twitter-account, This feed gets it´s input from the simplified mobile twitter interface. And now the E61 comes into play: It has WLAN, it has a webbrowser .. all you need to feed twitter (and thus all you need to blog a short message) while sitting in a keynote or walking around at the conference. This is Web 2.0 as i want it .. a simple way to connect services to produce content with several (very different) input devices.

I wrote many entries about Niagara2 in the past and will write much about the systems when i´m back home in Germany (in the meantime Allan Packers link list is a really god start). So i won´t write about the systems now. Nevertheless i want to write about the event itself. It was really cool launch. As far as i know it was the only launch (at last since i work for Sun) in front of 4000 peoples with at least half of them wearing "I love Solaris"-tshirts. Jonathan made an excellent keynote before and so it was a kind of a special atmosphere that surrounded the event.

The event started with the usual presentation about the advantages of the new systems hold by John Fowler. Then Andy B. and Rick Hetherington (lead of the N2 development) joined the stage and the real metal was presented. And this metal is really cool. 64 Threads. 64 Gigabytes. There was a Constellation-chassis full of T6230 blades. Doesn´t sound impressive.? This rack contains 48 blades. 3072 Gigabytes of memory in one rack. 3072 threads in one rack. I find this really cool. After testimonials from Ichiro Hirose of Fujitsu and Jason Turner of Mediastile, John talked about the six records on benchmarks (ranging from specjbb over specweb to specfp) and led to the Q&A session.

Andy B. made on of the best comments when he answered the question about the Oracle Licensin with "There are several opsn-source databases out there which runs really good on this systems". He got instant applause for this comment. I assume, that many colleagues share my opinion about Oracle ("incredible expensive, political pricing and unneeded for many databases").

So ... it was a really cool event. I think it would be a good idea to keep this way to announce such groundbreaking products. There is no better auditorium for such announcement than 4000 engineers. Maybe we can get even more ecstatic as the Apple folks when we announce Rock and Supernova The systems would deserve it ...

PS: I don´t know if i´m correct, but i thing we´ve offically announced Project Etude (Solaris 8 container on Solaris 10) yesterday on one slide. It´s called Solaris 8 Migration Assistant.

The CEC party starts in this minutes at the Palms in Vegas. I´ve decided not to go there, as i don´t like parties and - more important - i´m incredibly tired right now.

It wasn´t the presentation i´ve expected, but that was my own fault ... reading the abstract before the presentation avoid surprises. I had expected a presentation about Solaris RBAC. Nevertheless the presentation by John Walsh was quite interesting, as it looked to Role Based Access Control from a organisatorial perspective: How to implement Roles? Where are the challenges in doing so?

I took some interesting information out of this breakout.

• Often the first try of a customer to implement roles end in role explosion ( worstcase: vastly more roles than users)
• Don´t try to put 100% of all roles in your model. The project will never finish
• 80% of the people use 20% of the roles. 20% of the people use 80% of the roles.
• Based on this observation, define a standard set of roles for the 80% and use exceptions (together with a decent toolset) for the residual 20%
• There are two methods to determine roles: Top-Down (ask managers about roles) and bottum-up (ask authentication databases about roles). Mostly you end with using both methodologies in a hybrid approach. But whatever you choose to do, choose the methodology that has the least potential of role explosion

This was the only presentation i´ve left early. Nothing new in it for people who try to stay informed and the presentation had room for improvement (to say it polite) . Nevertheless i can´t write about the contents of this presentations, as it was marked as "Sun internal" only. Enough said....

I start the summary about the breakout sessions i´ve attended today with the best one. Tim Bray talked about his insights into Web 2.0. It wasn´t really about the hype inherent to this meme. It was more about the technological side of the phenomenon and foremost to development of such applications as the nature of the Web 2.0 mandates, that certain key issues are addressed by the development framework.

Tim Bray thinks, that there are for key issues to developers of Web 2.0: Scaleability, Time to market, Maintainability and Integration. Interestingly he doesn´t say "Java for all" becaus it´s simply not true. For example people opt for PHP because of Scaleability because of the default shared-nothing architecture of PHP application, and Time to market, as it´s easy to hack together an application. But the code is mostly unmaintainable, as Tim summarized it : "Bad SQL-Code, in Spagetti-PHP-Code in Spaghetti-HTML".

Ruby on Rails on the other side addresses Time-to-maket and maintainability, as code written for this framework tends to be more compact and by the fact that it´s difficult to evade unit testing. From the viewpoint of Tim, Ruby on Rails has the potential to get as big as Java is nowadays, once Ruby has solved it´s performance and scaleability problems.

Out of this reasons, Sun works together with the community: Warm relationships with the developers, sponsoring of equipment, employment of top developers for the JRuby implementation and so on. The talk of Tim was a good example of the new agnostic-ness of Sun. Java is a good answer to many questiions, but not for all. Almost at the end of his presentation, he came up with the tagline that summarizes this position: "The network is the computer. The computer is heterogenious. Deal with it!"

Tim was definitely the highlight of todays breakout sessions. I share his opinions.Programming languages are tools, not religions and we should handle them as tools: using the right tool for the given job.

When you look around, there is one thing even more ubiquitous than last year: The biten apple on the lid of notebooks. Looks like Sun engineers have a profound affinity to well designed technology even for their notebooks. And optically a MacBook Pro matches better with our server as a boring black notebook.