Design and Implementation of Picasso 2.0 and A Glimpse of the Future Version
Ravi Shetye · 2009
Picasso is a tool being developed in the Database Systems Lab for visually analyzing the behavior of industrial-strength relational query optimizers. Database Systems Lab had decided to come up with a new version of the tool, called “Picasso 2.0”. As an output of this project the lab released the tool “Picasso 2.0” in February 2009. This report talks about the design and implementation details of the new version. The new major features implemented in Picasso 2.0 are “Approximate Diagram Generation” and “Dimension Specific Range and Granularity Diagram Generation”. Also many minor features like “global coloring scheme”, “consistency in the slice displayed in different tabs”, “validity checks over the parameters the user selects”, “new command line syntax” etc. have been added. Efforts have been put in to make Picasso 2.0 bug free and more robust than its predecessor. This report talks mainly about the design and implementation of the new version of Picasso, “Picasso 2.0” and the bugs and loose ends fixed during the transition from the previous version to the new one. “Picasso 1.0” and “Picasso 2.0” provide various features which enable the user to generate diagrams that throw light on the functioning of a query optimizer. However the user has to manually explore the diagram to find areas with unexpected, irregular fluctuation in estimated cost or cardinality. Also analysis of various plan diagrams across the query-templates or across different database engines is not supported by current versions of Picasso. This report also talks about ‘Data Mining’ strategies proposed to be integrated in the future version of Picasso. These strategies would facilitate automated exploratory analysis to the user, thus enhancing the utility value of Picasso.