Evolving the Data Warehouse

From the Microsoft SQL Server Data Warehousing team

First post: Project Madison Questions

First post: Project Madison Questions

  • Comments 2
  • Likes

Welcome! To kick this blog off I’d like to share some of the most frequently asked questions regarding Project Madison, the new MPP data warehousing solution based on SQL Server.

Is Madison a Data Warehousing appliance? Is it the same thing as DATAllegro?

Madison will follow many of the same principles that DATAllegro employed in designing its DW appliance. Strictly speaking, Madison will be delivered as an appliance-like solution. Madison is different from the current DATAllegro product in terms of the hardware configuration, OS and database.  However, the vast majority of the “secret sauce” that enabled all of the hardware components and the software layer to work together will be carried forward in the Madison solution. Like DATAllegro, Madison relies on using industry standard hardware to keep costs low and facilitate scale-up and scale-out. The plan is to also keep the sales and support model very streamlined for the customer, so that they do not have to work with mulitple vendors to purchase or to log a support ticket.

When will it ship?

Our plans to release Madison in early 2010 are on track. Of course there is a lot of work to do between now and then. MTP and TAP beta programs will kick off this June and will run through early 2010.

What’s the feature parity with the DATAllegro v3 appliance?

Madison is planned to have functional parity with the last major release of DATAllegro, with a few critical enhancements to support some unique capabilities in SQL Server 2008. As we draw closer to MTP1 and the TAP programs, we will be able to share more details about the features.

Will Madison support SQL Server 2000, 2005?

Previous releases of SQL Server will not be supported. Madison is based on SQL Server 2008 Enterprise because this release has unique enhancements around data warehousing that result in a much more comprehensive and easy to manage data warehouse. 

Learn more about Project Madison on the home page and check out the Enterprise DW Architectures white papers on the TechNet portal for DW.

To those who have tuned in to the Data Beat blog, thanks! In the future, please join us here.

  • Joe, you wrote, "Madison is planned to have functional parity with the last major release of DATAllegro."  That's the "early 2010" release, which would be something like 2 years since the last major DATAllegro release.  So two questions:

    1) Teradata, Greenplum, Aster, Sybase IQ, Vertica, and others are not standing still.  There won't be ANY significant enhancements?  Not embedded analytics, parallelized or not? Not a columnar type?  Not support for large objects (esp. text), which I believe DATAllegro did not have?

    2) What about the non-functional side:

    query & load speed, support for heavy, mixed work-loads?  I hope you're going to do a lot better than DATAllegro parity (accounting for improved hardware) on that front.

    3) Will you really support scale-out?  Seems like the scalability model for Windows platforms is more naturally scale-up via more CPUs and cores.


  • Seth, great questions and I reckon there are other folks out there who are similarly curious. I queried our engineering and consulting team and we have some answers.

    1) There will be some enhancements, but our main focus is on porting the DATAllegro technology to the Microsoft platform. We’re confident that this will be very competitive overall.

    Consider that Dell recently unveiled its new Global Enterprise Data Warehouse on DATAllegro V3 technology (at TDWI Las Vegas). This is a pretty clear statement regarding viability of the current feature set for an EDW platform.

    We're not standing still either. The last major DATAllegro release ships in June of this year making it the second major release since the acquisition and the 3rd if you look back 6 months prior to acquisition.

    2) You are correct, recall that the comment in my post was specific to “feature” parity.

    3) Yes, Madison is a true scale-out technology, in the same way that DATAllegro was. Expansion relies on adding servers, not upgrading them (scale-up). This is consistent with the shared-nothing, parallel design and maintains the “appliance-like” provisioning aspect of the DATAllegro V3 appliance.

    The underlying SQL Server platform continues to improve in terms of scale-up and we’ll also be able to take full advantage of enhancements in that area.

    So, with this model we have the advantage of driving scalability improvements on both axes.

Your comment has been posted.   Close
Thank you, your comment requires moderation so it may take a while to appear.   Close
Leave a Comment