My photo
Auckland, New Zealand
Smurf sized geeky person with a penchant for IT, gaming, music and books. Half of industrial duo 'the craze jones'. Loves data, learning new things, teaching new things and being enthusiastic.

Tuesday, 15 September 2009

Gemini - new in SQL Server 2008 R2 (Kilimanjaro)

I'm currently at TechEd and thought it may be useful to blog session notes, both for myself as notes to refer back to, and just in case anyone else out there is interested in any of the topics. The notes are just that, notes... some of these I need to drill down in to when I get back to work to find out more information, but hopefully if this is a topic you are interested in the notes will give you a start point to drill down further as well. There may be some repetition of notes if I've picked up on an idea again later in the talk.

There are a lot of notes, so I'll split the blog posts by subject. This one is about project Gemini. This is the new self service BI tool that will be included with Kilimanjaro. These notes are combined from the Code Camp presentation on Sunday 13th and from Mark Souza's presentation at TechEd on the 14th.

Gemini Notes:
  • Slides should soon be available at (give Dave a few days to get these uploaded).
  • In-Memory Database.
  • Managed self service BI.
  • Accessed from within Excel 2010 after installing Gemini plugin.
  • Reports in other places, such as SAP? You subscribe to the report and it imports the report data into your Excel spreadsheet. After importing data, you set up a relatioship link between existing tables and newly imported tables.
  • Tables show as tabs in Excel.
  • Data Feed - similar to RSS Feed but with data. Uses same icon as RSS feed.
  • In-Memory Database allows 100's of millions of rows of data to be stored in Excel with minimal memory cost. This is due to storing in a columnar fashion and compressing the data. When stored in columnar fashion data can be substantially compressed unlike row storage, this is because columns are more likely to have repeated data that can be compressed to a single entity.
  • Speed of In-Memory Database is very impressive. Example used was applying a normal Excel column filter to reduce from 130 million rows to 2 million rows in less than 1 second.
  • Excel 2010 only shows columns that are in use - if you want more columns you add as required.
  • Can relate multiple data sources together at table level.
  • When importing data from various data sources you can use Excel filter technology to fine tune which data you want to bring across.
  • 101 million rows, including Excel overhead = 600mb.
  • Publish direct from Excel to Sharepoint.
  • Data in Sharepoint is no longer static. It can be refreshed either as a forced refresh or on a regular schedule as determined by the document owner or the IT maintenance peeps.
  • sumx/countx - new options that allow you to sum/count data from a tabular source based on an expression, i.e. =sumx(RELATEDTABLE(table),table[column]).
  • Multiple data sources importable into Excel, including but not limited to: Access, SQL Server, SSAS, Azure, Oracle, Teradata, Sybase, Informix etc...
  • Use 'slices' to create interactive data analysis app in Excel.
  • When published to Sharepoint, users don't need to have Excel installed on their local machine.
  • Sharepoint 2010 - theatre and carousel effects are very cool. :)
  • Gemini runs ''in process'' with Excel.
  • SSRS can link via a connection string to xlsx file that contains data.
  • Use MDX queries in SSRS and SSAS.
  • Sharepoint moves Excel data blob to SSAS.
  • When moved to Sharepoint, 100's or more users can now access the Excel data at the same time.
  • IT services or owner of doc can control the refresh rate of the data. This means we now have live data available in Sharepoint.
  • Once workbook published to SSAS via Sharepoint, it is now available to other tools, for example SSRS. In Sharepoint click on the document and in drop down select ''Edit in report builder".
  • Until release of SQL11, the Excel data blob will only be available as cube data that can be queried and not as a full blown cube that can be related to other cubes.
  • Things that are not there yet: Can't yet use Excel to create SSAS/Visual Studio projects as a start point for SSAS work. Can't yet create full-blown cubes.
  • Suggestion from Greg Low: Be a great idea when sitting with business users to use Excel as a prototype tool to figure out what they want. You then take your prototype query only cube back to project team to create fully featured SSAS project.

No comments: