New England Database Day 2009

By C L Kim | Posted on February 2nd, 2009

It’s Friday January 30, 9:25 A.M. as I start this blog.  Ever wondered what a roomful of 150 database uber aficionados get excited about? I’m about to find out while waiting for the delayed start of the 2nd New England Database Day at MIT’s Stata Center. Why am I even here? I had thought QuickBase might be too well kept a secret among database professionals, and had proposed to present a poster session on QuickBase architecture and customer use examples for this event. After encouragement and support from various QuickBase managers, I was further gratified to get personal interest to help from Jim Salem, QuickBase Architect, and Liz McCann, QuickBase Senior Marketing Manager; both provided the content for the poster.

I’ll do a separate blog on our poster; here, I thought I’d touch on some of the database research topics from the day’s speakers that might be of interest to the business technical users of QuickBase. The following are not meant to be summaries, but are just notes I took during the speakers’ presentations.

Prof. Mike Franklin, UC Berkeley and Truviso, Inc. “Continuous Analytics: Supercharging Query Performance with Stream Processing.” Not real-time query processing; stream processing. The research vision: classical “store-first” database is not ideal for business analytics of net-centric data. Want lower latency; but driver today is data volume growth. This is one of the two key-note topics for the conference and seems interesting to me in light of current interest in web user behaviorial analytics.

“Deep Web Search with Morpheus.” Example use case: How much is house at 44 Xxxxx Road, Manchester, NH, worth? Can use willow.com website; fill in form for site to generate information. Another example: I need a 3* hotel for less than $150 in Cambridge, MA. Can use hotels.com, and again ask question through form on web site. But data behind web forms is not visible to search engine. Morpheus — wrapper of user defined functions (that replicate a user filling in the web forms). The next web search frontier…?

Asst. Prof. Daniel Abadi, Yale University. “Data Management in the Cloud: Limitations and Opportunities.” (As an aside, Dan’s talk included an entertaining discourse on a post from The Database Column, a multi-author blog on database technology and innovation: ‘MapReduce – A major step backwards’ by Prof. David DeWitt and Prof. Michael Stonebraker.) If want milk, one can buy a cow, or buy bottled milk. Buying computer to host a database app is like buying a cow. Cloud Computing is like buying bottled milk. Data analysis applications more suited for cloud computing than transaction oriented applications. Food for thought…

David Karger, MIT CSAIL. “Baseless! Why the Best Database is No Database.” Proposal: Low performance database; object relational model; weak/no type checking; semi-structured data; simple queries only; direct manipulation; do not disclose existence of database; learn from user, not make user learn databases. “Exhibit“, a tool built at CSAIL for Database “Backed” Web Sites. Any topic on making databases easy to use naturally would grab my attention…

Alon Halevy, Google. “Structured data on the Web: where we are and where we can go.” Hypothesis: there are new opportunities for data management on the web if we focus on collaboration and lightweight tools. Topic: Deep-Web Crawl… three flavors. 1. Vertical search: a single domain; data integration techniques (e.g. Transformic, Morpheus); goal: close a transaction, or show related items, reviews, etc. 2. Search for anything; goal: drive traffic to relevant sites. 3. Product search; in-between above two. Topic: WebTables — a web-scale collection of tables; data is interesting, but there is much more in the structure itself: attribute correlation; synonym discovery. Topic: Organizing Query Results by Aspects: e.g. Kosmix; using dimensions to organize search results. This is the other key-note talk for the conference; always good to keep up with what Google is up to…