Archive for November 2008
If you didn’t make it to the last SQLBits (yeah, I know you had a good excuse: it was on a Saturday and you had to wash your hair, or your wife doesn’t let you out on weekends, or you live in Nebraska) then you’re in luck: a film crew from Microsoft filmed most of the sessions including the whole BI track and they’re now available for viewing on the SQLBits website:
I’ve already linked to Nigel Pendse’s initial comments on Gemini, which you can see here:
However those nice people at the OLAP Report have just given me access to see the subscriber-only content, which is probably the most detailed write-up available (I assume Nigel has been briefed by the Gemini dev team):
It doesn’t say anything much new, but I guess if you’re someone like Qlikview you probably want to get as many details on this as you can! Nigel is much more positive about it all than I was; he may well have a better idea of MS’s proposed solutions to the management problems everyone’s been highlighting: "Microsoft has some clear ideas about the role IT will be able to play in Gemini deployments, but the details were slightly fuzzy during the October announcements. It promises more details will emerge in the following months".
He also notes the problem of Excel users needing to define multidimensional calculations and says "Microsoft has not provided details of how this will be done, except to emphasize that Gemini users will not need to learn MDX. It’s proposal is to allow simple dimensional tests in Excel-like formulas in Gemini. These Excel-like calculation rules will be equivalent to MDX, but far easier for an Excel user to understand". Hmm, we’ve already seen MDX and PEL try and fail to make multi-dimensional calculations easy to use… I wonder if we’ll have a third stab at the problem?
Just had one of my occasional looks at Richard Tkachuk’s site, http://www.sqlserveranalysisservices.com and there’s some interesting new information on the home page. There’s an article on how to handle time intervals in AS (I’ve got some ideas on a different way of handling this, but I’d need to test them out to see how they perform), a note on how the Aggregate function works with calculated measures, and a draft of the AS 2008 Performance Guide that is a must-read:
Over on SQLIS.com, Allan Mitchell shows a variation on my original approach on how to build a cache-warmer in Integration Services:
The key difference is the Trace File Source component that allows you to connect direct to a trace file rather than have to persist the trace information to a SQL Server table. The component hasn’t been released yet but it will be soon (Allan and Darren have been working on it for a while); I’m going to try to use it in my PASS Presentation in two weeks. Once it’s out I intend to update my cache-warmer package and make it available for download from Codeplex or somewhere similar. I’ve got a few more ideas on how to improve it and I think it’s something that a lot of people would find useful.