Today was my keynote at Strangeloop 2010. My talk was first up and the title was "Enterprise NoSQL: Silver Bullet or Poison Pill". The session was well attended and there must have been at least 400/500 people there which was cool. I'd spoken with Bob Lozano the night before and I wanted to do a talk showing the mindset of the SQL crowd, the mindset of the NoSQL crowd, where the incapabilities or cultural differences were. That was the point of the talk and it seemed to work ok today. Lots of compliments after the talk and I got a good rating on the web site, a lot of talks didn't get any votes at all so I was pretty happy about it. I had hoped to leave more time for questions at the end but I only managed to get 2 or 3 in. I know it's something the attendees had complained about in other sessions.
I checked Twitter afterwards for tweets regarding the talk and the majority seemed very positive including one saying it was the best session of the conference so far. I'm just happy people got something out of it.
The only unhappy people today seemed be a couple of NoSQL advocates in attendance but you can't please everyone. The conference organizers videoed the talk so it should be online soon hopefully.
This was the first time I gave this talk so it can definitely use a little tweaking for the next time I give it.
First and most important, I attributed a spider man quote to Yoda. That was picked up straight away and led to many tweets about a possible lynching once the talk was over :) So, thats gotta get fixed. Whats funny is that I'd a premonition the night before and meant to google it to check I'd gotten the quote right. Next time, I need to listen to my 'inner voice'...
Next, Eric Bloch from Marklogic posted the only comment for the talk and it was negative. He doesn't rate me as an engineer because he thinks search should be done by an inverted index instead of map/reduce. Well, I guess I'll just have to disappoint him there, yes Mark, I do actually know what an inverted index is... Complex SQL like searches using typical group bys or aggregates would likely need still some map/reduce code running over a large dataset to get the results. That was my point. Frankly, I don't believe that strapping an inverted index engine (something like Lucene) on the side of a grid/nosql product gets you something as good as SQL for pushing out reports and the like. It takes more than just an inverted index to accomplish that. I didn't mean trivial searches, I meant proper SQL like queries in scope. But, whatever. I'll talk that as constructively as I can and refine the message next time to avoid such a miscommunication. Besides, I'd love to discuss over a drink some time if the travel gods allow.
Hopefully, people will still be excited by NoSQL like approaches but will also have their eyes wide open if they chose to go down that route. I tried to talk about the various things that seem to catch customers out when trying to use these new technologies. I threw in some race car analogies which I may pull but I thought the final line of "Driving a race car is a lot of fun but understand that without the right skill/training then you'll likely end up as the passenger of a high speed unguided missile very easily...".
I'd like to thank Alex and Bob for inviting me to the conference and hopefully it won't be the last time.
The slides are on slideshare but I think the video has a lot more information in it once it becomes available.
