Coming in early in the morning to the Dublin Convention Center (which looks like an awesome venue) and seeing the flood of people at the registration (which I avoided by being on time) I can tell that it’s going to be a big event. After the morning tea and before the opening I did the scouting to see the main conference room to see what companies have booths at the conference. Glad to see a large number of them and will return to them after they finish setting up.
The keynote series was introduced, as usual, but Jim Zemlin. He is a very interesting “CEO” (the Executive Director of The Linux Foundation) and I enjoy seeing him talk about Linux and the Linux Foundation. The first day of the conference matched Linux’s 24 birthday so we started with a happy birthday cheer. Well, to be fair, it’s hard to pin point an exact birthday (there is also Linus’s announcement on the Minix list on the 25th of August and the release of version 0.01 on the 17th of September) but the organizers were referring to the first public version, v0.02 on the 5th of October 1991. Also this week, is Free Software Foundation’s 30th birthday so greetings and thanks to them too.
Jim also announced new Linux Foundation projects, like FOSSology, a framework that scans for software licenses on projects to ensure license compliance.
Jim will remain the master of ceremonies throughout the conference.
The first actual keynote presentation was very interesting one, worthy of a TED presentation, under the catchy title “Man vs. Machine: High Frequency Trading and the Rise of the Algorithm”. The topic was basically Artificial Intelligence, but not the things that we are used to hearing for the last 20 years because of one reason: the AI we are used to think of as futuristic, is not only here, but has been here for a while and affects our lives daily. Center stage, stock exchange algorithms which are in control of the world financial markets. These came to be because of the limits of the human processing capabilities compared to the need for increased speeds in decision making. It’s both amazing and scary what that they currently do, but it’s important to see the limitation of them. And one of the focus is the things the algorithms can’t do yet (like “read” the news) but also what humans can’t do (respond fast) and why the human-AI interface is the next step in development of the world.
Next presentation was from IBM. I found out that the Power architecture was still alive and what IBM is doing for the Linux world. But it was an old school marketing presentation (though the presenter knew how to do her job well) and this is why IBM is the Diamond Partner of the conference. In short it was an advertisement for a line of Linux focused servers powered by..well… OpenPower. But the presentation was rather confusing and I still didn’t understand what was so open about OpenPower. I had to go to the IBM booth after the presentation to talk to some engineers to find out (stay tuned for another part of the article).
Last of the keynotes was about Drones. A popular geek toy these days but people usually are accustom to very simple drones which are just remote control flying cameras. However, the immediate future of them are actually autonomous drones. One of the presenters was from Willow Garage, a startup from Menlo Park that I actually interacted with about 4 years ago (I went on a volunteer study that involved testing one of their robots). They try to actually advance the world of robots in small increments. For example, they presented drones that could understand the environment and navigate in their surroundings independently. Slowly but surely, SciFi looking things are coming to the real world.
That marked the end of the first series of keynotes and the start of the about 12 parallel presentations in the smaller conference rooms.
The first such presentation that I’ve been to turned out to be awesome. I attended after a colleague introduced me to the term bufferbloat, which refers to the nasty latency problems that comes from the usage of buffer combinations in order to do network packet switching/routing. The presentation was titled “Bufferbloat 3.0: Recent Advances in Network Queuing” and presented by a cool guy from Brocade. When I walked into the room I saw on the stage an child-size inflatable pool, some plastic pipes and a series of liquid containers. I though to myself that “this presentation is going to be interesting”. And it was. The presenter was as entertaining as he was informative. He used the tubes and liquids to visually represent packet queuing algorithms. From FIFO to RED to combinations of queuing mechanisms stacked like a cake, he presented to pros and cons of them and when to use them and when not. The current state of buffering in the network world is not that great, but they people from the Bufferbloat community are trying to raise awareness of the need to think things though when it comes to handling packets on the Internet.
Next, I attended a presentation from Red Hat about SR-IOV Support in oVirt. Two years ago at my previous LinuxCon, I really enjoyed the oVirt presentation so I was optimistic. I wasn’t familiar with the term of SR-IOV, which stands for Single Root I/O Virtualization, a PCIe extension that allows a device to appear to be multiple separate physical PCIe devices. I wanted to find out more about the technical details, however, the presentation turned out to be more about oVirt configuration than the SR-IOV itself. Perhaps it would have been more interesting if I wasn’t familiar with oVirt. Still, I got the basic idea, which is that this offers hardware assistance to hypervisors when it comes to NICs.
Since this post is becoming longer than expected, I am going to end here and continue with a second part on the presentations, starting with the Kernel Panel from the evening of the first day.