The life of a CA Copenhagen Campus Connection

Lately I've been reading up on concurrency. 

I started out with Allen B. Downeys Little book of semaphores, went on to Edward A. Lee's classic article The problem with threads, which has the most fantastic language. Stumbled upon Brian Goetz talk from InfoQ on Concurrency: Past and Present and found a link to the golden article The free lunch is over: a fundamental turn toward concurrency in software by Herb Sutter, which depicts today computing environments.

Made a mental note on reading the chapter Beautiful Concurrency by Simon Peyton Jones in the Book Beautiful Code and concluded the brain walk with the new article in ACM Queue Real-World concurrency by Bryan Cantrill and Jeff Bonwick. But first read Bryans excellent blog-entry explaining the background for the article.

Unfortunately I don't have the answer to the classic question - so I leave the question up to you:
Is the free lunch over? Or is the buffet bigger and cheaper than ever?

Thursday I attended a High Computing Seminar at the Danish Technical University (DTU). Ruud van der Pas, Sun Microsystems was on his annual visit in Denmark and gave a talk on Throughput Performance on the UltraSPARC T2 processor, based on the notion "core equivalent".

If you havn't heard it - there's a new kid on the block after the Niagara 2 chip, and thats Victoria Falls - a 2-way T2+ chip - and this is a work horse! The more work you throw at it, the better it performs. Where the Niagara T2 chip has 8 processor cores each with 8 threads and each core has a decent dedicated floating point unit. The Victoria falls chip has either 2 or 4 chips, with 8 cores per chip, 128 or 256 threads in all. Also the 8 L2 cache banks has been freed up, to allow faster memory access.

Chip Multi Threading (CMT) is a technology that allows multiple threads (process) to simultaneously share a single computing resource, such as a core. This greatly increases the efficiency of usage of the core. At the same time, multiple cores share chip resources, such as memory controllers and caches, thereby improving their utilization. The result is unprecedented per-chip performance.

This is a really good video with David Patterson from UC Berkeley giving his views on the Niagara 2 chip performance. You propably recognize the name - he is the co-author of one of the best machine architecture books written to date. Patterson and Hennessy even based the "putting it all together chapter" of their book on the Niagara chip.

WOW - I can't remember a conference where I brought home so much cool stuff.

4 new shirts, a speakers gift (thanks COORD-team), a new bag (thanks Steve!), a new cap (thanks OpenSolaris advocacy-group for sending out user-group kits), a chokolate pumpkin (thanks Wim!) and of course my new Eee 901 PC (thanks GratisDNS).

And now back to the laundry pile accumulated from beeing away the whole week

Oh my - what a week! I am completely exhausted - but at the same time I feel hyper active and on top of that I caught the "Eee fever".

Its been a week of conferences and hard work and way too little sleep.

Wedensday I went to JAOO in Århus, where I was at Suns exhibition booth along with Thorbiörn Fritzon. I had the priviledge of meeting Guy Steele and watching him and Richard P. Gabriel give their very artistic keynote "50 in 50" - a tribute to programming languages.

Thursday I was back in Copenhagen running double time since it was the day before the Open Source Days conference and I had organized for Steve McKinty from the Open High Availability/Geo-team to give a talk on Open High Availability at the IT-University of Copenhagen in the afternoon. Steve gave me a REALLY nice present for this - a bran new bag filled with cool stuff!

Friday and Saturday it was time for Open Source Days 2008, where I gave a talk on SunSPOTs entitled "Computing beyond the keyboard, mouse and screen". About 50 people attended my talk and I think it turned out pretty good despite the fact that I was very nervous.

At the end of the conference GratisDNS gave away a bran new Eee 901 and I had also put my business card in their jar to participate in the draw - as the very LAST person. Peter Larsen from GratisDNS had stirred around in the jar right before Dan Klein chose the winner - nevertheless - it was my business card that got drawn!!! I thought it was a joke when Peter first meantioned my name - but it wasn't - I had won the Eee I usually never win anything, so I am pretty excited about my new black Eee 901! You might say I caught the Eee fever like everybody else at the conference

I even found a really cool blog entry about - how to run OpenSolaris 2008.05 on the Eee .....

MacOSX: 8 years of Innovation and Open Source

Jordan K. Hubbard - I've been bugging Jordan for ages (sure feels that way), to come to Denmark to speak about his work at Apple where he is combining his immense FreeBSD and Darwin knowledge. OS X is sharing a good deal of code with the FreeBSD project and Jordan was one of the original co-founders and core team members of FreeBSD and in my book that degree of knowledge makes someone worth listening to.

How to become a Puppet master

Have you ever heard of Puppet? Puppet is a mass scale systemadministration tool written in Ruby - its author Luke Kaines (Reductive Labs) used to be one of the core developers for the open source project CFengine. We use CFengine extensively at DIKU for the administration of our UNIX server park - but sometimes we get a little tired of the way CFengine does things. Puppet has a fresh view on things - and once in a while its really healthy to stop in the midst of all the "that should have been finished yesterday" - and contemplate whether things could be done with a little more cleverness - Puppet is one of those "wait a minute - thats nifty". While checking out Puppet I stumbled across a very interesting podcast from RedMonk featuring Luke Kaines and Nigel Kersten from Google. And I had hardly finished hearing it, before I was trying to figure out - how the hell I could get hold of Nigel and persuade him to come to Open Source Days - it turned out that Nigel is a big fan of Lego - and what better place to visit than Denmark Anyway Nigel's talk is for me considered a MUST.

One account to rule them all - OpenID

You know the feeling - you jump to a webpage, you need to download a piece of software, just to try it out, and you are yet again asked to create an account and enter your e-mail etc. *sigh* - yet another useless account. OpenID tries to rectify that by  lowering the user frustastration level and giving the control back to the user with a Singe-Sign-On OpenID account. OpenID is an open, decentralized, free framework for user-centric digital identity. OpenID is a young project, but it got a huge backing group in very short time, which makes its chance of survival pretty damn good. One of the companies that's begun using OpenID is Zyb - a service to keep an easy accessible on-line copy of the contacts from your phone. At Open Source Days, we have 2 OpenID speakers - one Friday morning and one Saturday afternoon, but on Saturday there will also be a lightning talk on the topic and a booth in the user group community area. Lots of possibilities to gain more insight to OpenID.

The once and future Roller

Another talk I am really looking forward to is developer Dave Johnsons talk on Apache Roller. Apache Roller is the software for mass scale blogging sites - Suns own blogs.sun.com is running on Roller and it's one of the worlds biggest corporate blogging sites with 5000+ bloggers. Apache Roller graduated from the Apache Incubator in March 2007 and is now considered an official Apache project. Roller was originally written by Dave - which means, that at Open Source Days you have a chance to hear about the past and especially the future of Roller from the guru himself! Apparently support for OpenID has just been added to Roller, thanks to Googles Summer of Code Apache Roller is a case story on open source development worth listening to!

My new favorite color? Chrome of course!

Do you want to hear one of Denmarks own Chrome/v8 cartoon heroes Kasper Lund? It's just been settled that Kasper will give a talk on Saturday October 4th on the development of Google Chrome/v8 - and the inner workings of the javascript engine that makes it so damn fast.

I hope to see you at Open Source Days - its always worth spending 2 days with like minded people and getting recharged with new ideas.

The registration for Open Source Days 2008 is now open! You can either buy a ticket that covers both Friday and Saturday or one for just Saturday.

If you can't or don't wish to buy a ticket - we are still looking for crew for the conference - that way you can get in for free and help make the conference even better.

I sure hope to see you at the conference - its the open source event of the year in Scandinavia. This year we've added lightning talks to the agenda - which means you'll have an even harder time deciding what talk to listen to - because there is an entire extra track!

Next week I'll participate in the Danish 5 km run "DHL stafet" along with a couple of employees from DIKU. That means I've been running for the last 2 months or so. Exercise is not really my "thing", but getting my new iPhone 3G made it a whole lot more fun.

I installed the application RunKeeper which tracks your runs - speed and GPS location. After a run you can login to the RunKeeper website and see your run. The application costs 59 DKK - which is one of the more pricey applications in App Store.

It has a little problem connecting to the GPS satelites, which means you start the application, and get a "determining location" - which takes a loooong time, right when you are standing there ready to run, and sometimes it even fails to connect and you have to retry - that's really annoying, but otherwise it works like a charm.

RunKeeper recommends having the iPhone on an armband out in plain sight, but with a combination of the danish weather and the iPhones ability to resist water - that is not really a good combo.

Anyway - wish me luck - it the first 5 km run I've ever participated in

Sun has officially chosen to support Software Freedom Day 2008 - on September 20th.

I guess that means, I'll be doing some SFD planning again this year .

It was a really nice announcement, to receive now that the CA program is starting up again after the summer holidays. My only problem - what should we do different this year? I've been doing SFD for 3 years (2004, 2005 and 2006) and I don't want to do the same thing all over again - fx. handing out CD's at the main street of Copenhagen - that was fun back in 2005 - handing out 800 CDs in under 2 hours.

Maybe I should organize a couple of inspirational open source talks?

If you have any ideas - let me know. 

"Hello - my name is Damian. I'm from the future"...with these words my Day 2 at YAPC::Europe 2008 started, when Damian Conway gave his keynote "A.B.C.D.E.F.G" about Contextual::Return in the morning. Btw - it was amazing what that module could actually do - does it brew coffee and tie my shoelaces as well?

Larry Wall - perl hacker extraordinare

At Day 1, I didn't manage to see more than the keynote by Larry Wall, because we were really busy handing out t-shirts at the registration desk, but at Day 2 - I attended 3 interesting talks appart from the morning keynote.

• Lars Jørgensen - Making Large Legacy Systems Beautiful

I know Lars from DIKU where we wrote a couple of projects together. He now works for the Wellcome Trust Sanger Institute in Cambridge UK, doing loads of perl hacking on their old legacy systems, and deciding on a daily basis what to recycle and what to refactor. The talk gave his personal view on how to handle legacy systems in a fairly big research environment. His final observations were pretty good, but he was a bit shy giving the presentation. Anyway it was good to see him again  

• Adam Kennedy - Tiny

Peter Makholm (I think) dragged me into this talk, telling me what a good speaker Adam Kennedy is - and I wasn't disappointed. But for a talk about tiny perl modules there was an enomous amount of big machinery involved!

• Léon Brocard - Working in the cloud

Léon is an amazing guy, having loads of modules on CPAN and doing loads of community work in London.pm and starting other Perl Monger groups. His talk was about cloud computing and how to put your business in the cloud and using the cloud services available to make your computing systems scalable. The talk was intended to be a discussion, but it was held in the biggest auditorium - and it seemed a bit hard to get people to interact in a BoF (birds of a feather) way.

This is what 300 lunchbags looks like

On Day 3 Damian started out by giving his second keynote "Beautiful Perl" and man you can do some serious ugly coding in Perl if you want to! On Day 3 I also heard 3 talks:

• Jonathan Rockway - Moose Introduction

I simply HAD to hear this one, because everywhere I went in the hallway people were saying Moose in every other sentence - and I had no idea what Moose was. I got wiser Appart from being an international greeting - its also a module making Perl more object-oriented and enabling a lot of the cool Perl 6 stuff into Perl 5.10.

• Phillippe Bruhat - Messing with git history using Git::FastExport

This talk was a bit dissapointing. The talker couldn't get his laptop to work with the projector and suddenly he was 15 min. behind schedule and unfortunately the talk got a little amputated by that.

• Chia-Liang Kao (CL) - Branch Management with SVK

So I stayed in the same room, after the git talk, to start the video-cam, but got stuck in a really interesting talk by CL, who works for Best Practical who also makes Request Tracker. The talk was about SVK - a tool to do offline editing in your subversion repository. Its also available as an Eclipse plugin. Boy that was a REALLY nice talk!

CL himself

I was really glad I finally got around to attending a YAPC - it was definately worth while. If not for the talks - then for the auction at the end. I've never before heard people paying 1.700 DKK to get Damian Conway to NOT rip up his book Perl Best Practices.... also don't be alarmed if you suddenly see the domain ihate.cpan.org  - it was auctioned off for a fair price

There's pictures on Flickr - search for yapceu2008 or ye2008.

I'll cross my fingers and hope I can get around to it - to go to Lisbon in 2009. The theme for next years YAPC::Europe is Corporate Perl.

So - the program for this years Open Source Days is shaping up really nicely. Hopefully we'll have most of it online next week. I am really proud of how many cool talks around Sun technology we've managed to fit in. 

• Why formats matter - keynote by Louis Suarez-Potts
• The once and future Roller by David Johnson 
  
• Open High Availability Cluster Geographic Edition talk by Steve McKinty
• OpenStorage/ZFS talk by Mads Toftum
• Deadly Sins of MySQL and PHP by Carsten Pedersen
• Beyond the keyboard, mouse and screen by myself (a SunSPOT talk)
  

Dave is also going to talk about Project SocialSite which has just been released and Louis Suarez-Potts is working on setting up a Nordic OpenOffice project leader/workshop meeting.

Todays qoute from YAPC: "The CBS wifi suck - oh look chocolate cake - the organizers are geniuses" - makes you wonder how much you can solve with a good cake...

I can't wait - its tomorrow! Time to relax, sit back and listen and talk to old friends I havn't seen in ages. The camera battery is fully charged, the laptop likewise. Its my first YAPC - so maybe I should start out by listening to the talk "How to get the most out of YAPC".

I have agreed to lend a helping hand to the organizers, so I don't know how many talks I will attend, but I am looking forward to the keynotes by Larry Wall and Damian Conway.

There'll be pre-registration beer @ the Globe tonight from around 18.00. 

See you in the hallway track

You've probably heard about skater parks before, but at Danfoss Universe in Denmark they've built Europes first Segway test track and for the next year, they'll let you ride it provided you give them feedback on how to improve the track. I couldn't resist - I HAD to try it - it sounded way too fun to pass up an opportunity like this.

Here I am trying to get over the small hills. Getting up hill was easy enough. Getting down was weird because you had to lean so much forward. If you lean backwards on the segway it drives backwards.

Here I am at the end of the track just getting the hang of it - see that big savy smile on my face?

If you can find the time to go to Danfoss Universe, its worth the long drive. Its a park dedicated to science and technology. In their new house called cumulus (its even shaped as a cloud) they have an interactive Bitland where you can manipulate technology in a physical manner, fx. you can text another "phone" but its so big you have to use your feet and jump from button to button.

You can also experience the Icelandic contribution to the World EXPO 2000 in Hanover.

Its worth a visit. 

At DIKU we have a very old print server (vaftrudner), that uses the LPRng project along with some really wierd homemade Postscript2-written stuff encapsulated in Perl, which should have died a long time ago, but didn't. Needless to say - its been on our wish list for quite some time, to set up a new print server based on CUPS (Common UNIX Printing System) and Kerberos for authentication.

Our computing environment has changed drastically over the past few years. Where as a couple of years ago all students would use the Sunray based terminal rooms, now days they all mostly bring their own laptop. So - what is the best way to provide printing services in this environment, where you have zero control over the students laptop? So far our solution has been to provide a CUPS-ssh backend (made by my collegue Martin Parm) which works on both OS X and Linux clients.

But CUPS provides support for Kerberos authentication from version 1.3 and the internet printing protocol (ipp) works on most platforms. Mac OS X, Linux and BSD is easily done, but Windows has for some time posed a problem. For some reason unknown to me Microsoft has only added ipp 1.0 support - but not ipp 1.1 support, and unfortunately the 0.1 version difference holds the authentication additions.

Fortunately a little piece of open source software gave us a solution to this problem: Redmon - A Redirection Monitor. Redmon install a new printer port on the Windows system, which can be redirected to another port and another program. So basically you install a new local printer, and instead of for instance a cable connecting the printer, you choose the redicretion port, which grabs the job and pipes it out through ssh. This is almost the same way our CUPS ssh-backend works today. If you have Kerberos, you can choose to use the Kerberized version of ssh and you can pass your credentials through ssh.

If you are a printer sysadmin, and provide a printing environment which includes a printing solution for Mac OS X, Linux, *BSD and Windows clients, I'd like to know how you set it up and if you have a better solution than the one I described above.

You know the feeling, you go scavenging among your old disks to see if there's something useable, but mostly the individual disks are too small. Its just a bunch of old disks.

JBOD rectifies that problem, by allowing you to use the disks as if they were one single volume. The proper term is spanning, but most people call it JBOD (Just a Bunch Of Disks) in contrast to RAID (Redundant Array of Inexpensive Disks). So wherein lies the difference between JBOD and RAID 0 (striping) you might ask. Well JBOD can be thought of, as the opposite as RAID - in a RAID you partition the disks, which usually chops them up into smaller pieces, in a JBOD you combine the disks into a larger logical volume. But neither RAID 0 nor JBOD provides any fault tolerance. In fact JBOD only has 2 advantages over RAID 0:

• Its avoiding drive waste
• Easier disaster recovery

Now enter ZFS into the equation, which is designed from the assumption that all disks will eventually fail. It has excellent data-loss detection mechanisms and it can even detect silent data corruption. With ZFS your disks are stored in pools and the mount points share the same pool, which means, that its harder to run out of space. You can always add more disks to the pool without having to reshuffle your data - that's neat!

There is a really great interview with the inventors of ZFS (Jeff Bonwick and Bill Moore) on the web. Its 45 min, watch it if you have the time.

There I was - in the server room, staring at the shiny new T5220 machine, which DIKU got through the Sun Try 'n Buy program ready to install Solaris 10, but my Macbook Pro had no serial interface. After a little googling around I found the PL2303 USB to serial driver (they are hosted on SourceForge), which worked as a charm.

I also downloaded Zterm, and a little while later the installation was well under way.

Now the machine is running MiG (Minimum Intrusion Grid)

Minimum intrusion Grid is an attempt to design a new platform for Grid computing which is driven by a stand-alone approach to Grid, rather than integration with existing systems. The goal of the MiG project is to provide Grid infrastructure where the requirements on users and resources alike is as small as possible (minimum intrusion). MiG strives for minimum intrusion but will seek to provide a feature rich and dependable Grid solution.