We’re happy to announce that Version 3.2.0 is now available of the MapD Analytics Platform.
MapD 3.2.0 represents a significant release in terms of MapD Core functionality. Of major note is that tables can now be sharded on a numeric or dictionary-encoded string key, which allows joins that can be sharded to scale to much higher cardinalities than before. Note that more efficient sharded group-by operations will come in a future release.
We’ve also added significant join support, including enabling multi-column joins (provided the combination generates a unique key), as well as one-to-many joins (i.e. joining a customer with all members in their household on household id). One-to-many joins on multiple keys will likely land in the next release (3.2.1).
We now allow joins on bigints, although it is still recommended for performance reasons to dictionary encode unique ids rather than storing them as integers.
Finally truncate table is now supported. You can expect more update and delete functionality to land in future releases.
The MapD Core rendering engine now supports different symbol types, for example allowing for hexagonal binning of heatmaps. Heatmap functionality will be landing soon in MapD Immerse.
On the MapD Charting Immerse side we’ve refactored our implementation of line and histogram charts, fixing many bugs and now allowing stacked histograms and range chart on histgorams. We also now allow users to lock numeric dimension axes as well as set the min and max for a dimension axis directly on the chart.
Please find the full release notes here. You can download the newest version here, or if you have already signed up reuse your existing link sent by email. We look forward to getting your feedback, and thanks again for using MapD!