Cognos Insight as a front end for TM1 (10.1) - Scalability

[whitej_d]

I've been trying to play around with using Cognos Insight as a front end for TM1 as a sort of dashboard, reporting interface. However I've been having some trouble with stability and speed. This is quite possibly due to working in a VM environment, and a whole host of other non-ideal aspects to the environment I'm working in.

As a matter of interest while I'm doign a proper test, has anyone had a go at using Insight as a front end to a large, calculation intensive TM1 model? Any feedback on performance vs Excel (or even Advisor) ?

[PlanningDev]

When you say scalability using Cognos Insight this brings up a couple of concepts.

Insight as a simple connected front end is going to suffer from the same issues as TM1 Web and the Excel Addin, namely WAN performance. I have noticed that insight's UI is a little slower than the others and as such larger data sets may take a little longer to render. I haven't noticed it being a whole lot different from TM1 Web. As far as calculation intensive, if you are in connected mode then all the calc's are being done by the server and insight is simply the UI. So in this case calculation performance is not different when using Insight vs the other methods.

The scalability piece of insight comes when you use it as a distributed client for Contributor Application. Depending on many model factors you may see an increase in performance or decrease in performance here. I would say 90%+ of the time you will see a performance gain. Really the dependencies here are based on the power of the client computer, IE RAM, and CPU since a "slice" of the contributor model has been delivered to then end user and in this mode all calculations are performed client side vs server side. The scalability gain here is that you have "distributed" the computing power across client machines rather than centralizing it on a server.

Basically, if you want to use Insight then your best experience will be if you have plenty of RAM and CPU available on the client machine.

One thing of note we have found is that most companies have real time scanning on client laptops/desktops. This causes performance drain for applications including Insight. In a lot of cases the servers can have real time scanning selectively disabled which helps quite a bit. You may want to check and see if this is an option for client machines as well.

Yes this was a bit long winded but in short, yes we do see some UI performance issues with Insight on laptops that don't have as much power as others.

[whitej_d]

Thanks, that's really useful feedback.

Interesting to hear about the effect of scanning applications. I'd heard of this issue with earlier versions of TM1 too, but I suppose it makes sense if Insight is running a psuedo-local version of TM1 on client machines that it would have an effect.

[Suharsh]

Hi,

It may look a pretty naive question. But since we are moving to TM1 10.1 from TM1952, I want to ask a pretty basic question.

What is the use of TM1 Architect, when the document says that Cognos Insight (TM1 10.1) is able to do data loads as well.

Is it just that Cognos Insight is a good looking brother of Architect along with some new features of viewing data (Cognos Insight = Architect + Data Analysis features)

Thanks

[dr.nybble]

Cognos Insight has an import wizard that builds TI processes. If you want to go beyond wizard-like imports and do a bunch of custom scripting then you either need to use TM1 Architect or TM1 Performance Modeler, as the full TI editor is not exposed in Cognos Insight.