kdb+ New York User Meeting – March 11th 2013

Speakers Included:

Dennis Shasha – Fun with Timeseries

Interesting fun talk on finding the highest correlated streams among thousands of streams extremely efficiently by comparing streams against random generated data rather than exhaustively against each other. Followed up by pattern detection over different time windows efficiently which I think is this paper: .

Aaron Davies – (dis-)functional select and de-queueing bugs

Slides available here.
To those that have ever been forced to write functional selects, the shortcut notation allows a much more readable form.

Joe Landman – Performance vs. IO & Memory etc.

A topic I find extremely interesting, unfortunately the material was probably too densely packed and I couldn’t follow along well. Hopefully the slides will be released so I can get the details at my own pace. Joe has a great blog http://scalability.org full of high performance articles.

Simon Garland – Kdb 3.0 3.1f

Multithreaded slaves as standard – start kdb with an overridden -s and you can run separate kdb processes on different parts that will each perform part of the query. Standard slaves with par.txt were proving less worthwhile especially for SSD’s. Curious to see how this will compare to mserve.
-23! – optimise reading entire tables by mapping them in all at once. Really I want this to happen automatically (query optimizer please) but I guess we have to be happy with the faster speed.

discussion panel: “Components, Frameworks and Nifty Internal Tools – The Good, Bad and Ugly?”

Dave Thomas, Bedarra
Ed Bierly
Igor Kantor, BAML
Nate McNamara, Morgan Stanley
Ryan Hamilton, Timestored
At times everyone seemed to agree their frameworks were good but others were questionable. Interesting to me was questioning from the audience on creating an open source library of kdb tools…
  1. No Comments