Additional follow up information from Peter Taylor
Thank you all for showing interest in the Big Data and GIS 'Show and Tell' on 20 November. We have received many requests for the presentation slides, which are attached. Denise McKenzie's (Global Geospatial Consortium) slides will be forwarded on when we receive them.
One question that came up in the 'Big Data' talk was whether the current solutions can work with SQL Server and we got the following responses after the meeting:
Pivotal: “Pivotal HD is a supported, scale-out Hadoop distribution with highly-optimised ANSI SQL capabilities and close integration with the Pivotal Greenplum massively parallel processing (MPP) RDBMS system. SQL Server access to both Pivotal HD & Greenplum is via ODBC/JDBC; bulk-data import & export are also supported”
I think this means SQLServer is not supported directly by Pivotal HD.
IBM: "Today BigInsights can natively use federation to Oracle, Teradata, Netezza (PDA) and DB2 based databases (including Federation Server). Through federation server we can reach all the sources indicated on the left on slide 5. I believe the current plan is to provide native SQL Server federation support in the release scheduled for 1Q 2015 ...Today using the product we can copy data from SQL Server and many other data sources using sqoop and jdbc to either copy the data into Hadoop or into a Hive table. Federation gives us the ability to query data in place at the source."
I think what this means is that BigInsights will link to SQLServer in Q1 2015 (subject to the usual IBM provisos about future products).
Microsoft: "Microsoft's 'Polybase' provides a single interface to bring together data from most popular Hadoop distributions (including Hortonworks, Cloudera, on premise and cloud) and SQL Server within Microsoft's SQL Server Analytics Platform System (APS). Polybase functionality will be extended to support other databases as sources including existing SQL Server databases (Scheduled for Summer FY15). Polybase is not available in the Software only edition of SQLServer and currently there are no publically announced plans to do so. However, existing SQLServer databases can be loaded/consolidated into a single APS for reporting where they appear as if they were the original SQLServer database and will perform significantly faster (100X). For more information, please contact James Cherry at Northdoor (mailto:firstname.lastname@example.org)."
I think what the Microsoft statement means is you have to buy APS in order to use Polybase over SQLServer in 2015 (and that will be by the summer).