6a00e3932f172e8834017d415d83ee970c-800wi

Day two of Strata Santa Clara turned out to be a day of sober
second thought. Actually, there was sign of that in day 1 with Tim’s session on
“why [we] can’t escape SQL”. But Kate Crawford’s keynote in the morning “Algorithmic
Illusions: Hidden Biases of Big Data” is another articulate counterpoint to the
boundless optimism surrounding Big Data. I am not a curmudgeon and I indeed
applaud Oreilly’s bravery to consider and run these sessions. When I see that there is finally some dissenting voice about the state of Hadoop, then truly one can see that Hadoop has grown up.

The highlight for me has to be Google’s presentation on F1. I
see F1 as a reasoned response to the BigTable-inspired technologies that are in
bloom today. F1 is a *gasp* SQL RDBMS with ACID properties which nonetheless
scales globally (literally). Even better, the slides are up already so have a
look to see the future.

I provided a small scale defense of why SQL and query
language matters at Intel’s booth in the mid-afternoon. I’ll share that with
you once Intel post-production is complete. To summarize, the trends that I
outlined yesterday regarding SQL hints at the future. Both Drill and
Stinger/Tez have allowances for new query languages beyond SQL (MDX?). I believe
this trend for extensibility shows a necessary maturing of the technology. Substituting
query languages with APIs works in limiting cases; for integration and continuity
with existing tools and constituencies, a query language—at least SQL—is necessary.