Search This Blog

Saturday, January 24, 2009

Pentaho - Open Source BI

Hi,

From past few days I was really very anxious to try my hands on Pentaho.

I heard a lot about the open source Business Intelligence Tools and two names that comes to my mind are Pentaho and Jaspersoft.

So I simply logged on to the Pentaho Website and looked for some demos. Then I looked at the hosted environment. The website is bit confusing and a new visitor like me found it really difficult to search for things may be because there are lot of things there.
Anyway, the good thing I found is this 'Try our Hosted Demo' which not many websites has.

The first screen after logging in looked pretty simple.

This is good. End users generally gets confused if you have so many buttons and they start loosing insterest from the first step itself. If the interface is simple, it gives them some sort of relief and confidence. The same is true with me.
Upper Left side pane shows the folder available and below it will show the files available in that.
The three icons in the middle of the center window provides you options to choose to create a simple tabular report(or a print quality report) or to create some charts. The third button is unclear.
Let's try on the first option first.

OK. This is a wizard. End users love wizards. If you can do something by simply clicking on Next button using defaults, nothing like that. so far so good.
Here I am not considering how the datasets are selected or from where the data is coming simply because the hosted demo environment does'nt have anything like that. So that's a seperate story and I will try to uncover that as well.
So selecting a Business Model, I think will be selecting a cube created for the purpose. That means we are doing OLAP. Now its not clear as of now what technology Pentaho uses, I mean ROLAP or MOLAP. That's a question as of now.
Let's move on and see what's there in Next.

that's a standard wizard. Available items and selected items. There is a distinct selection check box. I think here we do not have option to select which items we need to have distinct values. Checking this option will give distinct values for all items. Little confused here. No worry. let's move ahead.

Here we have an option to categorise the items under levels and put filters as well. Though I did not checked these options, I expect them to be standard.


Formatting options. Good. Users can do such things in the wizard itself. We have a huge list of paper options as well which can help in getting better print quality. You can preview the report in various standard formats. HTML, Pdf, Excel etc. Now the Next button has got disabled, that means we are all set to go and see some action now. Let's click on GO.

Here is my report. Not looking that bad. If I can get a report like this just by clicking on Next button, I am fine with that. But now I know I can make it much better. Interesting so far.
Lets move ahead and create an analysis report.
Simply click on the button having a chart icon on the top bar.

A small window appeared. choose the schema and the cube. I think this schema is similar to Essbase application which has multiple cubes inside it. The application generally is based on the functional aspect of an organisation like sales, finance and inventory etc..
For this demo we just have one schema and a cube inside it. Let's click on OK.

Now wait for a minute and take a look a it. Some buttons appeared on the top. In a way this is good to present the necessary buttons only. I mean when we were creating the static report, these buttons were not there and that saved a lot of confusion probably. Adding things could be easy but to make things clean and simple is really difficult. Let's appreciate this.
Let's click on all regions(+).


I generally like this kind of pivoting using +_ signs. Not every tool have this. I did not see this in Oracle OBIEE ans while talking to customers, I have observed that this is the most preferred way of having a pivot report. Also, one more thing I liked here is the ability to expand the departments for a particular region.

Now let's see what does these buttons means. The first button looking like a cube is OLAP Navigator. From this what I understand is you an add or modify the cube members. Click on it.

The icons are self explanatory but I did not understood there use until I tried them. So here you can change them to appear in rows and columns. By default, dimensions are under rows and measures under columns which is pretty logical. You can rearrange the order of their appearance and also apply filters. So I created a filter to have only one region.

Here is some difference. Here you can check or uncheck the members. But I did not see anything to select multiple members. I need to tick them individually to select. Here if we had that available and selected sections, it would have been much better. Because in actual scenarios, you have a huge list of members in your dimensions and in those cases it would be practically difficult to tick each member manually.
Also, I wanted to remove or hide Positions dimension from my report and failed to find any option to do that. Quite possible that I am not able to find it but if it not there, Pentao will earn one more negetive point.
Ok let's move ahead and look at the next option. The button says MDX. let's look at it.

It seems the database is Multidimentional but which one is not clear. Quite possible that Pentaho can connect to multidimensional sources directly. In that case, this would be very useful tool.
Other few options are very general like sorting and changing the layout, swapping rows and columns.
I am very eager to see some charts now. let's click on the chart button. I hope this would be some wizard based.

I clicked on it. no wizard but nothing less than that. A chart got created automatically. That's OK.
I actually wanted to see the options. There is another button besides the chart button. That is to edit the properties.

Options looking pretty OK. But I did not see anything in case I want to have the numbers or figures appearing on the bars or atleast on the chart against legends. Almost all standard chart types are available.

Two uncommon charts which I found are Pie charts by column and by rows. Not all the tools does have these charts. At the same time I did not see other advance charts like scatter charts. I am not sure if it has gauge charts and sliders. Quite possible that this may be available in some other version or component. One thing wrth mentioning is drill through capability. In properties only there is an option Enable drill through. I checked it. Lets see how does it work.

Here I selected pie charts by columns just to see how does it look. Charts are fine. Enabling drill through activated the hyperlink on the charts. clicking on the chart will create the drill through report.

This is fantastic. Generally creating drill through reports are not so easy. But this is something really easy and user friendly.
Now lets take a look at some sample dashboard.

Here on Dashboards, I see the values are displayed. But I am not sure whether these are text boxes created for the purposes or available in properties. The charts were also not interactive. I mean when I clicked on the region, though the sales trend chart below changed but didnt show any option to drill down. My guess is the first chart(Territory) is static with values pasted pasted on it. In that case the dashoard needs to be made more functional and interactive.
Another dashboard which attracted my attention is this one.

I dont know how this has been made or how difficult it is to create something like this but it's looking pretty interactive. On the map, if you click on any region, the values will pop up like this.

Overall, the tool is user friendly, easy to use, easy to understand, easy to manage and having good reporting, ad-hoc query, dashboard and OLAP analysis capabilities.

Though I did not checked its scheduling, notification and advance functionalities like row level security which may be out of the scope of this onlie demo. But I will try to gather all these information and probably will try to download the tool and loo at it from the scratch.

Any inputs from experts are welcome. Please spare me if you find anything incorrect and let me know.

Thursday, January 22, 2009

Gartner BI Magic Quadrant 2009

Yesterday just got a chance to look at Gartner's Magic Quadrant for Business Intelligence 2009.
I am really happy to see Qlikview almost touching the Leaders Quadrant Boundary and may be by next year Qlikview will join the leaders club.
So that's a good news for people who have invested in this technology.
This would be really interesting to watch how Qlikview will compete with the leaders. The strong points mentioned by Gartner are OK but I love to look at the weak points because thats the only area where you need to put more efforts and will eventually decide on your success.
1. Lack of statistical and predictive modelling: Thats the key area where Qlikview needs to improve heavily to be able to compete with other leaders. At present, Gartner named some of the competitors as Tibco and some other small products and while doing that it says it is behind them as well. If Qliktech fails to address this quickly before getting into leaders quadrant, it will become very difficult to move forward or maintain its position.
2. The fear of Qliktech being getting acquired will have some impact on the prospects. The prospects will get more cautious and may look for other options which are more certain and safe.
Prospects do not want to suffer from the change in policies, product names, re-architecturing etc and want to play safe wich is fair enough.
3. Gartner feels Qlikview still requires more examples of Large BI deployments and stressed on saying that it has not moved further in this area as compared to last year. I feel this will remain a challenge until Qlikview make some improvements in the architecture to deploy on large environment.
4. The last point may be very dangerous for Qliktech. People who were involved in large deployments understand the importance of Metadata management. making quick reports and good reporting capabilities are good but metadata management is the second pillar on which the deployment stands. If Qliktech fails to address this soon, it will definately be very difficult to get large deployment examples.

You may also want to read Qlikview vs Others which has some discussions on the pros and cons.

This was about Qlikview, the tool which I personally love.
The other interesting things which Gartner mentioned is inclusion of some open source BI tools like Jaspersoft and Pentaho. I hear a lot about Pehtaho and would love to include a review for this in my blog soon but before that I would like to try my hands on that or read some technology information whitepapers.
Now with the inclusion of open source BI, these tools will get some acknowledgement and people will have a choice to look at them as well.
Other open source which I am hearing a lot is Jaspersoft.

Another thing which Gartners mentioned is the SaaS(Software as a Service) BI tools. This may be good for products based on some properitery technology to store the data which has a potential risk of migrating the complete application if customer chooses to shift to a new technology or if the vendor plans to de support or does not provide a way to integrate with other technologies. In that case, customer has absolutely no choice other than to competely rebuild the entire application on different platform which I think will not be an easy and economical task.
I really dont have any idea how SAAS vendors make sure customers investment is not affected if anything of such sort happens. If someone can provide an insight would be helpful.
The new names which I never heard are Pivotlink, Lucid Era and Oco.
to be continued....