As expected, being a stand-alone conference made it difficult to attract what I call “casual attendees”. Folks who would be happy to spend a day or two after a big conference but won’t take the time out of their busy schedules for inter-continental travels just to attend this symposium. Consequently, most attendees were either authors themselves or locals happy to spend an extra hour in the morning rush hour to get to San Jose.
Anyways, the conference was well worth the time — even though I’d liked to have seen more database/data management related talks.
One of the highlights was definitely Scott Schenker’s keynote on Software-defined Networks (SDN). Making networking sound interesting and appealing to a larger crowd is a feat. Luckily, Schenker’s immensely talented and able to break down a complex technical subject to a fairly diverse crowd that, let’s face it, is just not that much into networking in the first place. The talk very quickly turned into a software engineering lecture that underlined the often forgotten essence of a good architecture: proper abstraction and conceptual independence of different layers. And we’re not talking about the data plane with its networking layers but the control pane — that’s what the network community apparently overlooked in the past and finally fixed with SDN’s: clean abstraction and hierarchical composition of components. All summed up nicely in a quote from Liskov’s Turing Award lecture: “Get the right abstraction, that’s how things are done”. Layering the architecture allowed rev’ing each layer at its own pace without having to rewrite the entire stack. As a result the adoption of SDN was phenomenally fast and successful — compared to, say, IPv6.
The actual sessions were much in the spirit of what SoCC embodies: a hotchpotch spanning networking, Operating Systems, and lots of VM provisioning. Organizers did a great job reconciling the different objectives and ended up with a fairly strong program if somewhat light on the data management aspects of things. Which could be due to one of two things: (1) database folks aren’t good at writing papers and all database submissions to SoCC were just not cutting it, or rather (2) the Cloud though interesting is actually not really that interesting to database researchers; mind you, these are the same folks who missed the MapReduce train a few years ago.
Lots of people floated interesting ideas how to go forward with SoCC. The most persistent theme in the mix of suggestions was to take it local, i.e., have it take place in Silicon Valley every year and turn it into one of the fixtures of system conferences that the area is famous for already. Definitely has my vote; I’m looking forward to SoCC’13!