Cool I tell you!

Chris, Corrigan and I went to Google in 2008 for MeetBSD. We had a great time, so when I saw the announcement earlier this year for MeetBSD 2010, I rallied the troops.

The event was Friday and Saturday, and it took place at the Hacker Dojo in Mountain View. The format of the conference was the same that Puppet Camp had, where the talks were proposed and arranged by what the attendees wanted to talk about. So, the conference started off with James from iX Systems putting the Google Document URL up on a board, and then he let everyone vote and propose topics

Our main MC was James Nixon. The flow was great, and I think he did a very nice job and getting people to participate and tease out details from everyone. He did that by having everyone stand up and gather around after a break-out session was over. Then, he had everyone talk about their break-out session. As much as people probably hate being put on the spot, it really tore down some social barriers. I probably would have been a more passive participant, but a format like this helped me/forced me to be more engaging. It was excellent.

So, that was the first day, and I got to hear about CLANG and LLVM and what it means to the BSD community, and how the new LGPL license effects the base system (for example, no GCC 4 in the base of FreeBSD, it will have to be a Port). I got to hear Matt Dillon talk about in details his HAMMER File System (B-Tree file system with sweet rollback support), SSD drives, Hybrid Drives (and why they are not a good idea), and why ZFS and HAMMERFS will not outperform UFS in the use-case of a Database. Oh, and why Log devices are so critical for ZFS and HAMMER FS. I also got to tackle Pawel, the ZFS developer, and ask him about the roadmap for v28.

A few other really cool talks:

• David Maxwell – The State of The NetBSD Foundation. This was very fun and informative, he presented some of the details of NetBSD’s development and funding, while asking FreeBSD, OpenBSD and DragonflyBSD members to talk about their own. It was a great way to learn about the BSD’s and their difference. It was also a well done presentation because it was reciprocal and respectful off the variants. Really good.
• John Kanen Flowers – Kane|Box. John has a strong presense, and I am interested in seeing where Kane|Box goes. Since it is both a commercial hardware device, and an open source project, I may run it on my server at home.

I also spent some time randomly going up to people like Dru Lavigne, Pawel Dawidek, and Will Backman (of BSDTalk, one of my regular and favorite podcasts), just to tell them I appreciate what they do. I’ve always taken such a passive role in conferences like these, yet I’m so enthusiastic about the BSD’s, that I have to recognize that I AM a part of the community. I don’t develop, but I do blog, document, and talk as much as I can about them. Lets also not forget, I’m a happy end-user, being a SysAdmin of FreeBSD systems is pretty fun. The point is, I’m trying to put myself out there more, be more involved and participate in the discussion.

With FreeBSD reporting their largest status report this quarter, and the excitement everyone here seem to have, I think it confirms the solid and steady progress of the BSD’s. One person did comment to Chris, Corrigan and I about the age group at the conference. He didn’t say that he thought it was good or bad, but he said that the people in the BSD community tend to be “more mature”. I think he would like to see younger people more involved, and I agree. I’ve always felt that the BSD’s are driven from an engineering perspective. They don’t have marketing department telling them they need to include support for the latest buzz-word. This is good to me, otherwise, you see ridiculous things like clustering packages that are completely virtualized. However, the negative aspect to that is the lack of media attention. I sometimes thing it would slightly benefit the projects if people in the academic world saw it as ‘sexy’.

Thankfully I didn’t have to get up at the crack of dawn (7:30!) because it was scheduled to start at 11am. I was a little rushed for time since I had to drop Caralyne off at my parents in Knightsen (so I back-tracked a bit), and I left at 9am, so I barely made it in time for the ZFS talk.

A Closer Look at the ZFS File System

by Pawel Jukab Dawidek

I’ve heard a great talk on ZFS from Bill Moore, one of the primary developers from Sun, but this talk was really cool because it didn’t just say what ZFS is from Sun’s marketting department’s point of view, but the technical details its mail peices and how it integrates in FreeBSD. He also discussed the current status, and we should soon see the ZFS version get bumped from 6 to 13! Pawel went a little into why ZFS just doesnt work on a 32bit machine, and i guess Sun figured that by the time ZFS was out, everyone would be phasing out 32bit machines in favor of 64bit ones :) Wishful thinking, thankfully my little server is a 64bit system and handles my ZFS /home volume just fine.

There was some mention of ZFS’s limitations with high-volume databases. Hopefully ZFS matures quickly, it is hard to go from UFS2 which has 20 or so years of QA, but ZFS has so many new cool bells and whistles that PJD said it most appropriately with “ZFS will do for storage what VM has done for memory”

Summer of Code

by Murray Stokely

Murray, a Google employee (who helped organize all this and yet I forgot her name, shame on me), two students and a NetBSD developer/mentor talked about Google’s Summer of Code project. I don’t have much to say on this except I think its great that Google does the Summer of Code project for so many open source projects, especially with ones like FreeBSD that have very few corporations willing to pay someone full time to be a commiter.

FreeBSD Foundation Update

by Robert Watson

Robert helps run the non-profit side of FreeBSD that takes care of donations, legal snafu’s, and getting things like Java certification for FreeBSD, manage project grants, and helps developers attend important dev. summits and conferences. Since FreeBSD is 100% voluntary, donations are crucial in the post dot-com bust. A few companies, like NetApp, Isilon, Ironport, Juniper… have made large donations, thats because they all use FreeBSD in their commercial products (and with the BSD license you can do that). For the FreeBSD Foundation to keep its 403(c) status, it also needs at least 1000 individual donations, and it doesn’t have to be big, just a lot of people.

It’s been about 4 years since I last donated, and I decided it was time to pony up again. I refuse to purchase Windows ( my more recent copy was free because I did a survey!), but I use FreeBSD in all the important aspect of my personal IT life, and I love to see it in the server room at work so a few donations here and there is the very least I can do for the OS that I personally beleive is of the utmost quality.

Crypto Acceleration

by Phiilp Paeps

I’ve always had a fascination with hardware crypto accelerators. Its mainly because I’m always concerned with two things: security on the wire, and performance. Hardware devices just seem to be the natural solution!

What I took away from this talk was an answer to that, and its “sometimes”. A lot of crypto accelerators only work with a limited set of algorithms and key sizes, and when you application doesn’t fall into that it falls back to the cpu. There was also some interesting facts about why a 32bit pci crypto accelerator isn’t so useful on a 64bit architecture. It was very cool to hear and I’d still like to get my hands on a nice HIFN card

Seamless Service Migration with PF and FreeBSD Jails

by Josh Paetzel

I wish I knew PF more so I could do this. He has a fantastic solution to upgrading a network service seamlessly using PF and Jails. Yes, you could setup a virtual machine, or just have a redundant box, but he was able to do this with 1 FreeBSD machine and it’s native tools.

I’ve setup a few jails at work before VMWare was the hip thing (and before ESX took off) and it amazed me how powerful that OS level of virtualization can be. I setup a full training environment, which was one server that had 8 virtual FreeBSD server’s for the students to trash and play around with. I also setup a FTP server in a jail to protect the host OS.

What Josh did here was very creative and cool, and maybe I can sit down in the future to implement it so my Apache and MySQL upgrades can be a little more robust.

Isilon and FreeBSD

by Zach Loafman

Isilon builds cluster file system appliances that are built off of FreeBSD. FreeBSD kicks complete ass when it comes to any network service, and with a rock solid file system like UFS it makes perfect sense to build and sell a turn-key solution off of it. That is what NetApp has based its business off of.

Its nice to see a company like that give back to the FreeBSD project, they did a lot of NFSv4 improvements that the base system could use.

Help! My System is Slow! – Profiling Tools, Tips and Tricks

by Kris Kennaway

Its hard to teach someone how to troubleshoot, there are no hard and fast rules, but Kris did a killer job at starting off with some simple tools like top(1), and figuring out where your bottlenecks are. Aside from a few specific kernel parameters a FreeBSD admin could tweak, this presentation translates to anyone who is trying to speedup their server.

He also went a little into benchmarking, and both profiling and benchmarking went over some very useful and pragmatic steps to quantify “performance”.

Conclusion

I had a great time this weekend, and I would easily give up more another weekend to do it again. I got to spend some time with Corrigan and Chris and simply “Geekout” on FreeBSD with a collection of cool and very polite people. I had a few misgivings about the structure of the convention. First, Michele hated the mascot, and I can see why. Having that mascot instead of the traditional Beastie or the new official FreeBSD Logo prevented me from sending out this MeetBSD link to other System Administrators that I work with. It doesn’t offend me, but it might offend someone else, or even worse, make them think this is not a mature environment. I know, the fact that people may be offended by a mascot alone is the opposite of maturity, but its a fact that people WILL be offended by it. I also thought the whole laptop+projector issue could have been streamlined by having ONE laptop properly configured with the projector, and then have the speakers run their presentation off of a usb drive.

Other than those things, the people who set this all up are awesome and I can’t thank them or Google enough. So go Google for FreeBSD and give them both some credit. Better yet, support FreeBSD with a donation at http://www.freebsdfoundation.org/

I spent the weekend at Google for MeetBSD in celebration of FreeBSD’s 15 birthday. I drove 70.2×4 miles, on a weekend, and for Saturday I got up slightly earlier than I would have for work. All worth it, this was the coolest mini-conference I’ve been to. Of course, the last conference I went to was BSDCon in 2003, and that was nice as well. The point is, I don’t get out all that much when it comes to conferences, I just make an exception for my favorite UNIX OS, FreeBSD.

When I got there for the registration (9am) I stood in line next to two gentlemen from Sweden, Karl and Pontus (or Pontis, didn’t know for sure). I call them gentlemen because they were EXTREMELY polite; so polite I was audacious enough to assume they were not from the US :) So, I chatted with them for a bit, picked up my swag-bag and waited outside for Chris and Corrigan to arrive. The swag bag was the most impressive one I’ve seen so far (even more than what Splunk usually provides). It contained a shirt, shot glass, hip flask, coffee mug, mouse pad, cool FreeBSD propaganda, a mouse pad, and a 2GB flash drive! The shot glass and hip flask were little hints that those involved like to party a little, and while I didn’t attend the after party, people were talking about it the next day.

Corrigan arrived first, then Chris, and we sat down just in time for the first speaker.

FreeBSD Network Stack Performance – Optimizations for Modern Hardware

by Robert Watson

Probably the best talk of the weekend. Robert Watson is really good speaker, and his presentation was well laid out and was just at my level. He did start off saying this talk wasn’t for kernel hackers or developers. It was really cool to see how things like TCP Offloading in hardware effects the network stack, the kernel, and how it can cause some complications with certain kernel level utilities like TCPDump, firewalls and packet filters.

Isolating Cluster Jobs for Predictability and Performance

by Brooks Davis

Brooks had an interesting presentation, and I’ve always wanted to see FreeBSD in the cluster market even though I know Linux has won that one out by far. I did feel the presentation was a little unfocused, or maybe i misinterpreted the title to mean something else. It was still interesting to see how you could use various methods to allow engineers different tiers of a cluster. His FreeBSD cluster of 1400 cores is controlled by Sun Grid Engine.

Embedding FreeBSD

by M. Warner Losh

I liked this one, and M. Warner Losh was a good presenter. Having him walk through how he prototypes an embedded device, the tools used to strip it down, and then getting it to boot in 2 seconds was very impressive. He also had a very good sense of humor throughout the conference.

PC-BSD 7 from a Developers Perspective

by Kris Moore

PC-BSD is geared towards the Desktop/Laptop market, and has a corporate backing. I can easily compare it to Ubuntu or Redhat Enterprise Linux. Its still FreeBSD, no arguing that, and you can still do all the cool things on it like use the Ports system, but it has a new and efficient package system like YUM+RPM or Ubuntu’s Synaptic/APT system. This is crucial in my mind for a desktop environment, especially laptops, where you don’t want to build everything from the Ports tree (which is all source code), you want a standard and consistent system, where you won’t always be building a custom web server. It was a good talk, but one thing that sort of drives me crazy is the name. See, its PC-BSD, but really FreeBSD, so in my kind of environment at work where we tightly control what is on the network; how are we supposed to classify this system? As PC-BSD, or FreeBSD? If someone asks for a FreeBSD desktop, do was say they are using PC-BSD or FreeBSD? Thats just my thing with it. Aside from that, I really welcome a corporate aspect to FreeBSD.

BSD Certification

by Dru Lavigne

There is now a certification process for FreeBSD. It’s new, and right now there is the BSDA certificate which is aimed at Jr. System Administrators. It covers FreeBSD, NetBSD, and OpenBSD, and the explanation of what went into the test, the Psychometrics involved, and the awesome price point of $75 for the exam. This was so it would be widely affordable to everyone across the globe. I thought of taking the test, but I was having too much fun and I didn’t want to feel that pre-test anxiety.

Xen Virtualization on FreeBSD

by Kip Macy

I’ve had some exposure to Xen with Redhat, so its nice to see it on FreeBSD, and it was really a quick status update on the project. Kip then went into something completely different and I didn’t take any notes on it so it’s already lost in my brain :)

From here the PC-BSD team brought out “BSDGirl”, which was the equivalent of a booth babe. After a few minutes of her parading around I turned to Chris and Corrigan and talked about how I felt about it. See, I hate to judge her, I mean, she may really be into FreeBSD and likes to dress up (and she does seem to have a web presence pimping out PC-BSD). It was the fact that she seemed to isolate herself from everyone there, and didn’t seem to talk to anyone or network with the folks there. I just felt the event was cheapened with a marketing stunt like that and Chris and Corrigan felt exactly the same way. I don’t know, that’s the kind of crap I expect from big companies that don’t care, not the small BSD community that I take personally. It didn’t seem like anyone else there was all that interested in her, so it was even more out of place.

We hung out until the Google staff pretty much kicked us out, and Chris and Corrigan went to the after party for a little bit. I headed back home, watched a little TV and went to bed so I was ready for day 2.