PerformancePoint First Impressions
I was meant to be at the third day of the UK PerformancePoint airlift today, but because of the snow we’ve been having here in the south of England I decided against the trip to Reading and stayed at home to build a snowman instead. However, I thought I’d go ahead and blog about what I learnt on the first two days; Charlie Maitland was/is also there and has also been blogging too, here and here. It seems like the PerformancePoint hype cycle is getting up to speed – there’s certainly a lot about it in the press, eg at twdi http://www.tdwi.org/News/display.aspx?ID=8313 - so I might as well add my thoughts…
First of all, what is PerformancePoint anyway? It’s really two separate things: first of all there’s the monitoring and analytics stuff, which is partly the next version of Business Scorecard Manager, partly the server-based dashboard, analytics etc components that came with the Proclarity acquisition; there’s also the budgeting and planning, ‘business performance management’, application which was codenamed Biz#. Initially I was a bit confused about why all this was being bundled into one product: I can see that the people using the bpm functionality will want to use the monitoring and analytics functionality, but the vast majority of the people using the latter will be people upgrading from the old Proclarity components (or moving away from equivalents like Panorama) and won’t have much direct interest in the bpm side of things. However I suspect that this is another case of Microsoft doing some clever bundling. The companies who are getting their dashboards/scorecards for a reasonable price (and the price is very reasonable, $20000 per server plus $195 per CAL) will suddenly find they’ve got this bpm application thrown in and decide to give it a try, in the same way that many companies came to Analysis Services because they got it free with SQL Server.
The course itself was almost all to do with the bpm functionality, with only a few demos of the dashboards and scorecards, so for the rest of this post I’ll just be talking about the bpm functionality. So, then, the key question… is it any good? Do we like it? Of course it’s impossible to say how good or bad a product is going to be based on an early CTP, but in general I was quite impressed. A lot of thought has obviously gone into it and it will definitely be a better option that trying to build your own financial application using Analysis Services. I wish I had some experience with other similar products, especially some of the better-established AS-based competitors like Outlooksoft, to be able to make a comparison but I don’t; I doubt that most people who evaluate or go on to use PerformancePoint will be judging it against its competitors though and I’m sure that as a V1.0 it’s missing a lot of the features that they have.
Down sides? At the moment the UI is a bit of a disaster, more VB4 than Office 2007 in terms of look and feel and usability, and a lot of work is going to have to be done before the September release in this area. There’s also, in my opinion, a tendency to use parent/child hierarchies rather too much (though admittedly many hierarchies in financial applications have to be modelled as parent/childs) and I would be worried about the performance implications of this although the program managers present at the course did say that the AS dev team had implemented some changes in SP2 which will mitigate this. Kristian Wedberg also made some valid points about the lack of support for integrating existing cubes and dimensions into a new PerformancePoint application which I fully agree with: at the moment you have to build every dimension measure group from scratch but realistically if you already have an AS-based application you’re going to want to reuse dimensions and measures from it, probably using linked objects.
There’s also something of a tension apparent in the design, stemming from the intent that non-technical users should be able to do as much of the work of creating and maintaining a PerformancePoint application as possible. If I recall this was always what Nigel Pendse liked about Applix TM1 in his reviews in the OLAP Report, so it’s a valid goal, but as always there are some things that are best left to technical people. The parent/child hierarchy issue is a case in point: after I had raised the question I had a good discussion from one of the guys there from Solver, a partner who has been doing a lot of work with PerformancePoint, and he said that to get around the problems associated with large parent/child hierarchies they’d found that they needed to break up the dimensions into smaller pieces and only add the relevant pieces to models/cubes. Now this is a perfectly valid answer and indeed PerformancePoint makes it easy to do just this, but it’s something that a business user should never be expected to know about or understand despite the fact that dimension design is a task that the UI wants to make easy for business users. Similarly, there’s the option to populate your dimensions either manually by typing in member names or by loading data from a csv file, both of which can be done by a non-technical user through the UI, or by pushing the data directly into the staging db using something like SSIS, clearly a technical person’s task and by far the most robust option. My feeling is that the division of tasks between those that can be done by a non-technical user and those that can’t could have been made better and that if, say, the model design tasks had been implemented in Visual Studio à la BIDS for the techies and just the maintenance and day-to-day management had been left for the non-techies in a separate UI, the product would have been stronger. But I understand that’s a contentious opinion and that this feeling is based on a very limited knowledge of the product and financial applications in general, so I might change my mind…
I don’t want to dwell on the negatives too much anyway, which I know I’m inclined to do, because as I say my overall impression was favourable and we’re still about six months from RTM. Before going on the training I thought that the bpm side of PerformancePoint would be something I might steer clear of in my day-to-day work, but now I feel enthused to go out there and do something with it. There’s a healthy amount of MDX used in its calculations and reporting so that has to represent a business opportunity for me!