It seems an age and a half ago, but the NLUUG spring conference on File Systems and Storage was just over two weeks ago, on may 7th. As the (now ex-) head of the programme committee I’d like to take a moment to look back and sum up the conference and what I got out of working on an event like this.
The NLUUG was the Dutch national UNIX users association — like many national UUGs in Europe, founded about 25 years ago — and has now re-branded itself for Open Systems and Open Standards. Those are more important in the long term than the proprietary UNIXes and provide us with a broader range of topics to think about. The NLUUG has been organizing two conferences a year for over ten years; this is the stated core business of the association. Some topics come back regularly, like security, whenever the state of the art has changed. Most of our conferences try to combine practical, sysadmin-oriented talks with cutting-edge research or development talks.
This year, File Systems and Storage, is a fairly low-level UNIXy topic. Now, I’m a BSD guy, so I think UFS should be good enough for everyone, but that is taking a narrow view of the whole. For one thing, it neglects the storage side of things — mostly I remember wrestling with vinum about eight years ago and being relieved when GEOM showed up.
Still, Linux has grown three new file systems in the last two kernel releases (possibly more). Those address different ideas about disk organization and new notions about metadata on files. Solaris got ZFS, which (re)combines the storage pool with filesystem management. [[ I should note that when I’m not a BSD dude, I do OpenSolaris things. Taking stock yesterday, I realized I have Free Software operating systems on most of my machines and two popular proprietary operating systems (I should cut that in half) but nary a Linux kernel in sight. ]]
And then there’s clustered file systems and file systems for virtualized environments (where de-duplication becomes important). The breadth of relevant problem areas and the range of technical solutions is huge. And now I’ve still neglected storage, with provisioning, mirroring, deduplication, backup and whatnot.
Suffice to say that the range of topics is pretty broad.
Unfortunately, we didn’t get a comprehensive range of file system talks; only four if I remember correctly. Several Samba-related talks, libferris (Plan 9 on steroids: everything is a filesystem), storage architectures and some search and semantics (after all, what’s the point of storage if you can’t find stuff?) rounded out the programme.
The conference itself picked up 285 attendees, which is a little below our target number of 300 for an “ideal conference” — but given the economic situation this is not all that surprising. Attendance also varies considerably depending on the topic. Security and virtualization are perennial favorites.
The day opened with Ted Ts’o, CTO of the Linux Foundation and maintainer of the ext4 file system for Linux. Later in the day he did an ext4 talk, but the opening was, surprisingly and most interestingly, an economic one. The message came down to this: SSD’s are not going to solve all our problems. In a whirlwind exploration of the economics of storage and the hard drive business, Ted showed that with current — or even next-generaton — processes it’s impossible to replace hard drives. The SSD remains a cool bit of kit, though. It was a great opening keynote; speakers later in the day referred to it in their own talks, which I think shows how much of an impact it had on everyone.
Since I was session chair all day (without my signature horse whip; I reserve that one for Akademy) I didn’t even peek in on most of the talks (we have three tracks), but the comments I heard from attendees were overwhelmingly positive.
All in all a good conference, and I’d like to thank our organizing bureau and the other PC members Ralf van Dooren, Melanie Rieback and Jos Vos for a constructive and interesting season of cooperation. Now I’m looking forward to upcoming conferences again — like Akademy, as part of the Gran Canaria Desktop Summit, and the NLUUG Fall Conference, which has “the Open Web” as topic.