The MS gemini team posted a new blog with a overview of Gemini, what was new to me was the following part:
Sharing Gemini Applications
While many workbooks are built for personal use, some are worthy of being shared across a workgroup. Here again, Gemini works the way Office users do. Since Gemini data is stored within an Excel document file, any way to move that document – through file shares, emails, publishing to SharePoint, etc. – transport the Gemini contents along as well. Users without the Gemini addin can browse the data, those with the addin get the full experience. Just as Excel and Gemini light up together, Gemini also extends SharePoint capabilities in several ways.
For the more visually-inclined amongst us, a flat SharePoint list leaves something to be desired. File names, data last updated and by who are useful but only tell part of the story. Gemini provides Silverlight based skins that present different views on document libraries. These views show snapshots of the contents of documents. In the example below, we see two workbooks with two spreadsheets within them:
These snapshots are also live links in that clicking on a thumbnail of the a worksheet will take users directly into ECS with the worksheet loaded.
Scheduled Data Refresh
The Gemini model embedded within the spreadsheet keeps information about where data came from. Once published to SharePoint, users can specify schedules for the data refresh operation so the workbooks use the resources of the server to stay fresh.
And what was even more impressive:
Using Gemini Applications as Data Sources
Once published to SharePoint, Gemini models embedded within workbooks appear as an Analysis Services databases! This means any AS client tool – Excel, Report Builder, etc. – can connect to this database as if it were on just another AS server. The only difference for these clients is use of a URL to the document stored in SharePoint instead of a server name. Gemini services running on SharePoint handle loading the right database, managing its lifetime, and transparently redirecting client queries to the right database on the right server.
This gives some incredible new options, but i wonder what it will take of server perfomance (memory ?) when you have 20 of these models on your server.
Read the entire preview here:
after posting the question about the performance on the original msdn blog post i got an answer from one of the Gemini team members:
You’re correct, the Gemini embedded data engine loads data into memory. However this is only while the models are in use. Gemini’s SharePoint services manage the lifetime of these models and move these in and out of the SP content database transparently from the end users, the only realization users might have is the first time they connect to a model it takes slightly longer because we’re extracting the workbook then extracting the AS database and loading it up in memory.