Can Dynamic Cube be used in Tableau?

[cognostechie]

Hi All -

I created only one dynamic cube so far and that was more than an year ago so couldn't do this research myself.

Since the dimensions and facts are stored in tables in a database and since Tableau can use a database as a source, does anyone know if a Dynamic Cube can be used in Tableau to make a Dashboard? Anyone tried it?

Thanks

[MFGF]

Hi All -

I created only one dynamic cube so far and that was more than an year ago so couldn't do this research myself.

Since the dimensions and facts are stored in tables in a database and since Tableau can use a database as a source, does anyone know if a Dynamic Cube can be used in Tableau to make a Dashboard? Anyone tried it?

Thanks

Hi,

A Dynamic Cube is a technology that runs over star/snowflake schemas in a data warehouse and includes in-memory caches and utilises in-database aggregate tables to make access to the data much more performant than if you were reporting purely over the underlying tables. In theory, any reporting tool can be used to report directly off the tables themselves, but only Cognos Analytics can utilise the technology that Dynamic Cubes leverages. Since Cognos Analytics can create much more detailed and controllable reports than Tableau, and can also create dashboards that look just as good (and are also far easier to create), why would you want to go anywhere near Tableau?

MF.

[sdf]

1 year working with tableau, for me I can say that tableau is not cube friendly.
It pains me when one of the useful use of tableau (data extract) is not available for cube source.

[MFGF]

1 year working with tableau, for me I can say that tableau is not cube friendly.
It pains me when one of the useful use of tableau (data extract) is not available for cube source.

With a Dynamic Cube, the data lives in your data warehouse, and the benefit is (and it's a HUGE one) that you know the data is governed, reliable and trustworthy. Using Cognos to present this data also brings in analytics governance - if your CA has been secured with any thought to common sense, you know users cannot see/do things they should not be able to. Putting both of these together, this means any analytics you deliver are showing results that can be trusted, with traceable lineage back to where the values come from, and audit capabilities that prove you haven't messed with the numbers. As soon as you start sucking out the data and putting it somewhere else, that governance and confidence in the data evaporates. Yes, I'm sure that Tableau dashboard looks really pretty, but what real value does it serve if you don't know if you can trust what it is telling you?

Just my tuppence :)

MF.

[cognostechie]

Hi,

A Dynamic Cube is a technology that runs over star/snowflake schemas in a data warehouse and includes in-memory caches and utilises in-database aggregate tables to make access to the data much more performant than if you were reporting purely over the underlying tables. In theory, any reporting tool can be used to report directly off the tables themselves, but only Cognos Analytics can utilise the technology that Dynamic Cubes leverages. Since Cognos Analytics can create much more detailed and controllable reports than Tableau, and can also create dashboards that look just as good (and are also far easier to create), why would you want to go anywhere near Tableau?

MF.

It's not me, it's the market. Most companies using Cognos are still using ver 10 and the visualizations in those are no match to Tableau and they have already brought Tableau licenses. Ver 11 also earned the reputation of being a 'buggy' version and that's my opinion too. It takes a while to convince people to switch and also depends on the budget available. Till then I don't want my sales to suffer so thinking of all options.

Thanks for the answer. It gave me what I wanted to know.

[cognostechie]

With a Dynamic Cube, the data lives in your data warehouse, and the benefit is (and it's a HUGE one) that you know the data is governed, reliable and trustworthy. Using Cognos to present this data also brings in analytics governance - if your CA has been secured with any thought to common sense, you know users cannot see/do things they should not be able to. Putting both of these together, this means any analytics you deliver are showing results that can be trusted, with traceable lineage back to where the values come from, and audit capabilities that prove you haven't messed with the numbers. As soon as you start sucking out the data and putting it somewhere else, that governance and confidence in the data evaporates. Yes, I'm sure that Tableau dashboard looks really pretty, but what real value does it serve if you don't know if you can trust what it is telling you?

Just my tuppence :)

MF.

Doesn't Tableau allow setting security? Data governance depends on the technical team. Even Cognos has different layers (FM, Transformer, Dynamic Cubes) in which the security has to be set . It does not automatically flow from the Data Warehouse.

[MFGF]

Doesn't Tableau allow setting security?

Yes, but this is entirely defined by the author of the dashboard. There is no concept of there being an overall security framework that applies to authors as well as consumers. It's not just a security issue, though. It's a capabilities issue too. A Tableau author has the ability to do everything, and there is no ability (as far as I am aware) to restrict certain capabilities from certain groups or roles as you can in a properly governed solution.

Data governance depends on the technical team. Even Cognos has different layers (FM, Transformer, Dynamic Cubes) in which the security has to be set . It does not automatically flow from the Data Warehouse.

It certainly starts with the technical team, but it has to be pervasive across the organization. As soon as you start extracting data for use in desktop solutions, you lose that governance. You can argue it's the same with Excel, and that Cognos allows you to dump out data in Excel format. I agree - it does - but the key issue here is that as an organization you have the power to prevent this in Cognos if you wish to. It again comes back to analytics governance - something Tableau doesn't have, as far as I'm aware.

Does security automatically flow from the data warehouse? If you have set it up in there, based on individual logins or groups, then you can configure Cognos to log in to the database under the relevant ID for the user or group. In this case the security certainly would flow through. Often security isn't defined here, though, and there is a need to define it centrally for your BI solution. You can do this in FM/Cube Designer etc, meaning your Cognos authors and consumers only see what they ought to. I'm not a Tableau expert, but my understanding of the way their security filters work is that they are defined by the author of the dashboard, not centrally as you would in CA. This means you have no method of governing what data authors see, and the implementation of the security rules can vary from author to author.

Anyhow, we've drifted way off topic here :) An interesting discussion nonetheless.

Cheers!

MF.

[bdbits]

MFGF, what you are saying is only true if all you have is Tableau desktop. Most organizations of any size have Tableau Server (or its cloud equivalent), which is a whole different ballgame. My current employer has the on-premise Tableau Server.

Its not quite as fine-grained as Cognos but Tableau Server has a full permissions model, with capabilities and authorization permissions as you would expect out of any tool aspiring to be enterprise-wide. Security is most definitely not up to the author, in fact at my current employer authors have no ability to set permissions. There are permissions for creating, editing, and consuming content, and read/write permissions to projects/sub-projects (roughly equivalent to Cognos folders). Functionally this is not much different than Cognos. It is a bit different in implementation of course, and Cognos is more fine-grained in capabilities. The folder structure in Cognos is better than Tableau projects/sub-projects in both management and user presentation (in my opinion). But the security is actually not that much different than with most any enterprise-oriented BI/analytics tool.

Tableau is not a separate ungoverned data store, in fact data governance is a regular topic at the Tableau conferences. With the desktop client, you do have an option to pull the data into a workbook if you want to, and the result is very much like an Active Report or using Excel with embedded data. Also similar to Cognos, Tableau does have its own technology you can use to pull in data to the Tableau server (extracts), much like you would a Transformer cube. But Tableau can also directly read a relational database in real time, and many other data sources. At the server level, this is analogous to a Cognos package. You can pass user credentials through to the data source if you want, or use a server sign-on. Tableau does not have the equivalent of a Cognos security filter embedded in a package, though; you are correct in that being something defined in the workbook. This is quite annoying and we are trying to find a good workaround, but are probably going to have to resort to stored procedure code (I think) as a workbook-embedded filter is a weak solution at best.

Tableau is great at visualizations and dashboards/stories, but falls pretty short when it comes to standard textual type reports (e.g. most financials). Their sales people will sometimes even tell you this. It is very, very limited in that respect. Also, I think for large organizations Cognos has a better semantic layer, and is better at a wider variety of reporting scenarios. But it all comes at a price that is quite a bit more than Tableau, so if you can live with its weaknesses Tableau can be an alternative. And it does render some pretty rainbows and shiny ponies, even a unicorn now and then.  ;-)

I still think Cognos is pretty great, but Tableau is a lot more than the desktop client these days.

[cognostechie]

Great information bdbits ! Thank you !

Yes, I know that Tableau is not an alternative to the entire Cognos suite because it does not have reporting capability and also does not allow making any cubes though it does allow creating hierarchies in a way. For reporting including modelling, many companies are buying Power BI these days.

But it all comes at a price that is quite a bit more than Tableau, so if you can live with its weaknesses Tableau can be an alternative.

The price depends on what deal you get. As a business partner, if we were to re-sell it at the same price we get from IBM then Cognos (admin + explorer) would be cheaper then what Tableau offers for Server + desktop. However, I am comparing my price from IBM with Tableau's MSRP and that may not be a fair comparison so won't press on this point but I don't think price is the reason why people use Tableau because whoever uses Tableau also has to buy some other license for reporting so Tableau is not an alternative but an add-on from a cost perspective.

Considering the various features Cognos offers which Tableau doesn't, a stabilized CA (which has the ability to let the users make visualizations easily) will probably be the best in the market but in absence of that, it's hard to convince customers. A stabilized CA would have a huge advantage because the customers wont have to buy another tool which is necessary with Tableau.

I hope this happens soon because as I see here in US, Cognos market is drying up, replaced with a combination of Tableau and Power BI.

[bdbits]

Tableau Server has what they call the hyper engine. Basically, it extracts your data source result set and stores it in an internal format that I suspect is a columnar data store and is quite speedy to execute, but rather slow to refresh. We seldom use this feature, but apparently a lot of people do as it is always a topic at conferences and Tableau is tweaking it a lot.

Tableau can produce tabular-style reports. It's just not very good at it, and you have little control. I am not sure they are particularly interested in improving it, either.

PowerBI is getting lots of traction these days. It gets pushed at our business people rather frequently by consultants and vendors. I still think it really only works very well if you are all Microsoft, and their efforts to extend beyond that (Power BI Server) are weak, in my opinion.

I find your pricing discussion interesting. I will say Tableau will discount pretty heavily from MSRP, at least they did for us and we are relatively small. I do feel Cognos could come back in a big way if they would fix all the annoying bugs and as you say, stabilize the product. And proper marketing; I think IBM could do better there if they wanted to. The product has a lot going for it functionally, but the glitz and buzz is around the Tableaus and PowerBIs primarily due to a better marketing message. Sadly, over my decades in IT I have seen marketing win over technical excellence far too many times. Solidify the product and give it flashier packaging and they could gain back some traction beyond the large IBM-centric shops.

Boy, we have really wandered off-topic now, haven't we?  :D

[sdf]

Tableau Server has what they call the hyper engine. Basically, it extracts your data source result set and stores it in an internal format that I suspect is a columnar data store and is quite speedy to execute, but rather slow to refresh. We seldom use this feature, but apparently a lot of people do as it is always a topic at conferences and Tableau is tweaking it a lot.

Very true! Most clients I had before when I was still a consultant uses Tableau primarily for (ad-hoc) dashboard/metrics/kpi seldom rely on it for reporting purposes. This thread suddenly turned into a pre-sales cheat sheet. :D

[cognostechie]

I do feel Cognos could come back in a big way if they would fix all the annoying bugs and as you say, stabilize the product. And proper marketing; I think IBM could do better there if they wanted to. The product has a lot going for it functionally, but the glitz and buzz is around the Tableaus and PowerBIs primarily due to a better marketing message. Sadly, over my decades in IT I have seen marketing win over technical excellence far too many times.

Another major factor which drives the decision is the value of the product that is visible. The superior functionality of Cognos (macros, parameter maps, advanced security, drill downs to name a few) exist but the users don't know that they exist because their developers have elementary knowledge of Cognos and tend to live with that as long as they are getting their paychecks. What is delivered to the users is a solution that does not have any better features than what any competitor offers. Since the 'visible' difference between Cognos and Power BI or any other tool is zero so there is no advantage in having Cognos. I have worked for companies where 250 copies were made of the Flash report because the developer didn't know how to set a data filter so he made 250 folders in Cognos connection, saved one copy in each folder and set the security on the folders so that every salesrep will see only his/her data. When a column was to be added to the report then the developer had to add that to all 250 reports (extra time resulting in extra cost). In another company, the database layer from FM was published in the package without making a Business Layer. Result - no self-service BI ability so the BI department makes every single report for the business. Moreover, for each report, a separate view is created in the DW because he doesn't know how to create joins in FM without running into data issues. The 'counts' are done in the views and then aggregated in the reports resulting in counts being wrong at the sub-total and grand total levels. He doesn't know how to use count distinct in FM. Result - extra time and cost + user dissatisfaction because they don't get a self-service BI front end. Now with this, when another vendor approaches them and demonstrates a self-service BI ability then they will jump for it because in their eyes, Cognos does not offer a self-service BI ability! Result - New product sold and Cognos out !  'Visible difference' is what matters rather than the 'actual difference' between tools so a weaker product gets better traction.

Sadly, this is the case with most companies and management is unable to understand that the BI managers/directors are unqualified or are there just because they know someone there ! 

I have started a practice to overcome this. I am letting companies know that whenever they get an answer from their developers that 'something is not possible' or 'this is what the current technology allows' or they cannot deliver the same way as the users want then they can consult my company and we won't charge to give them the advice of whether it can be done or not and if it can be done then how to accomplish it. That way, they would know what they can get. If they want to go further and have us implement it, they we will charge. I think if IBM offers this to all of their existing customers, it can result in a huge success.  The problem is not in the technology (other than what IBM is already overcoming) but in the visibility !

Yes, we are way off-topic  ;)

         

[Reinhard]

For a few months now I keep reading about Tableau as replacement of Cognos. Now I do get the better dashboards and the independence from IT which as the previous post shows can be crippling. Additinally in Tableau you just hand your employee the software and a datasource and he can start while in Cognos you need a professional for the server installation, the modelling, the security concept.....

So the cost difference is not really in the licensing. Nevermind though, why is Tableau so public at the moment? One of my customers at the moment is trying to replace 10000 Cognos licenses with Tableau ones.  Why is Tableau so ahead of PowerBI? No one mentions Qlikview or any other competition.

Btw, best discussion ever.

[MFGF]

For a few months now I keep reading about Tableau as replacement of Cognos. Now I do get the better dashboards and the independence from IT which as the previous post shows can be crippling. Additinally in Tableau you just hand your employee the software and a datasource and he can start while in Cognos you need a professional for the server installation, the modelling, the security concept.....

So the cost difference is not really in the licensing. Nevermind though, why is Tableau so public at the moment? One of my customers at the moment is trying to replace 10000 Cognos licenses with Tableau ones.  Why is Tableau so ahead of PowerBI? No one mentions Qlikview or any other competition.

Btw, best discussion ever.

I'm fascinated by the idea that is put forward that Tableau provides self-service BI for business users. Again, I'll hold up my hands here and admit I'm far from being any sort of expert on Tableau, but what I've seen leads me to believe that the success or failure of an implementation depends entirely on one or more power users who create the dashboards. I would not class these people as business users, just as I wouldn't class someone who creates FM models or creates standard reports in Cognos as business users. If you need power users to create dashboards in Tableau, which are then consumed by "regular" users, this doesn't sound like self-service BI to me? I would class self-service BI as giving "regular" users the ability to do their own discovery, and for this the process needs to be simple and intuitive. I think this is what IBM are trying to deliver with the dashboards piece in CA. Now before you say "but hold on, you need some data first", yes - that's true, and it comes back to the points I was making earlier about the need to have data you can trust. Somebody somewhere needs to prepare that data to make it suitable for non-technical users, and that's true no matter what BI tool you look at. But what interests me the most is the idea that a non-technical user can develop a dashboard on their own, without having to get a power user to build it for them. I'm very happy IBM are following this approach with CA dashboards, because it is as close as you can get to self-service BI for regular business users. Does Tableau really offer this? Happy to be corrected, but I'm not so sure it can... :)

Deep in the mire of off-topicness now, and getting deeper with every post :)

Cheers!

MF.

[cognostechie]

Additinally in Tableau you just hand your employee the software and a datasource and he can start while in Cognos you need a professional for the server installation, the modelling, the security concept.....

Yes, I do believe that the major factor is that Cognos requires quite a bit for technical work while Tableau does not but how is it possible for you to make Dashboards without any kind of modelling/mapping? I am curious to know what kind of datasource do you have? If it simply connects to a database and exposes the tables and fields then those are likely to have technical names without any relationship set between the tables. How do you determine which field(s) and tables(s) to use for your Dashboards?

You would know what you want but you wouldn't know where to get it from without the front end being intuitive with names that mean something to you and with relationship set between multiple tables. What am I missing here?
   

[bdbits]

cognostechie - I do believe your stories on what you run across, but wow. And what a shame, which sadly mostly points at the incompetence of the IT staff behind those installations.  :'(

As far as Tableau, here is the thing. There is a significant portion of the Tableau user base that is started and is still based on the desktop product. I am not sure they even offered server initially. Many of these users come from statistical and analytical backgrounds, and do not consume a data warehouse. They have access to raw data, and are knowledgeable enough to mash multiple data sources together within Tableau Desktop (basically in-tool modeling). Tableau has even recently released a tool that is basically an ETL product to produce extracts for Tableau. Other tools do the same kind of thing - model it inside the software, and download the entire result set. Inefficient? Error-prone? Multiple versions of the truth? Yes, certainly can be a problem.

But Tableau has a lot more enterprise customers now, and the issues of data integrity and trust are being surfaced (BI history all over again). These same enterprise customers are often pointing Tableau at data warehouses now, and things are at the point where you can use Tableau with Server more like a more traditional BI tool - consuming a curated data warehouse, using a semantic layer for presentation, on a web interface, with data filtered for security, etc. But the do-it-yourself community is also still there and going strong, so the impression one gets about Tableau can be very much affected by the source of your information about it.

Tableau has some interesting and useful capabilities. It can produce more than dashboards, and can make useful stuff (well, not so much textual reports but I digress). It does not require 'power users' to produce output if you have them consuming Tableau data sources (essentially a semantic and security layer). There is a web-based interface instead of desktop and you can control user capabilities if needed. It can be very much a drag-n-drop operation for regular users, and as self-servicing as any tool when implemented well. It is fair to say it is not as well-rounded as Cognos, and its strengths are largely in interactive graphically-based analytics - and you have the abilities for hierarchies and drill up-down-through and viewing raw data and all the rest if needed. Sometimes, this is all that is needed and it works fine. Not as capable as Cognos, but it fits a certain use case.

The cost of BI/reporting has often centered on licensing costs. I would argue the real investment is in data discovery, quality, and cleansing, consolidation, and curation on a well-designed data architecture. Ideally, you build a repository of data accessible by any capable tool, whether it is Cognos, Tableau, PowerBI, Qlikview, or whatever, even multiple tools if you like. That is how you deliver value from a data perspective, then implement your front-end tools well to engage users and reach the nirvana of self-service BI. That's my take, anyway.

[cognostechie]

Very informative. Thanks bdbits !

Let's see what we can do to repair the damage done already. The capabilities of C11 and somebody making a proper intuitive package for every source system would probably be an answer.

[Reinhard]

You are 100% right! The weakness of tableau is, that you need a well modeled DWH because you basically have BI users who need to join tables and propably have no idea of the data structure.

However, earlier that year Tableau Prep was released what is the Framework Manager of Tableau.

[bdbits]

If you are using Tableau Server, users do not need to join tables. In fact there is no UI for them to do so. Nor does it require a data warehouse, well modeled or otherwise.

Tableau Desktop is used to create the Tableau equivalent of a Framework Manager package. You pull in the data you need, create any filters, etc. then publish it as a "data source" which server users use as the basis for their reports like you would a Cognos package.

Tableau Prep is more comparable to an ETL tool, and is used to create Tableau data extracts.

[sdf]

Isn't there a "Link datasource" in tableau, I remember this form tableau 8?
Something done for  data blending.
Which seems to me is the interface for joining.

And I heard form tableau 10. There's this cross-database joins. Well this is for diff db sources.

[bdbits]

When talking Tableau, there are some differences between desktop and server UIs and capabilities. Blending is one of these. You have to use desktop. You cannot create data sources on Tableau Server, you can only consume the ones that have been published there by a desktop user.

[Reinhard]

If you are using Tableau Server, users do not need to join tables. In fact there is no UI for them to do so. Nor does it require a data warehouse, well modeled or otherwise.

Tableau Desktop is used to create the Tableau equivalent of a Framework Manager package. You pull in the data you need, create any filters, etc. then publish it as a "data source" which server users use as the basis for their reports like you would a Cognos package.

Tableau Prep is more comparable to an ETL tool, and is used to create Tableau data extracts.

Thanks, I mistook the meaning of a datasource in Tableau.