+ All Categories
Home > Documents > Qlikview or Oracle Obiee

Qlikview or Oracle Obiee

Date post: 15-Oct-2014
Category:
Upload: faisal
View: 1,058 times
Download: 1 times
Share this document with a friend
20
QLIKVIEW OR ORACLE OBIEE A Quik Look at QlikView Outside the world of Oracle, one of the rising stars of the BI world has been Qliktech, the Swedish software company behind QlikView . QlikView is one of a new breed of desktop, in-memory BI tools (see also the forthcoming Gemini tool from Microsoft, an in-memory, column-store BI tool that’ll be delivered as part of Microsoft Office 2010) that are being evaluated by customers as alternatives to traditional relational query tools such as Oracle BI Enterprise Edition and OLAP servers such as Essbase or Oracle OLAP. So what is QlikView, and how does it compare to Oracle’s BI tools? QlikView, compared to Oracle’s relational and OLAP query tools, has a slightly different approach to storing and querying data. Whilst it can load data from relational and file-based data sources, it then stores the data in-memory in a compressed form with associations defined between data items rather than the traditional joins that we see in relational query tools. The idea behind these associations is that they are derived by Qlikview during the data load based on matching up columns across tables with the same name, and then when you query the resulting dataset you can analyze your data using these associations. To take an example, the following QlikView demonstration shows data for a consulting company in a Windows thick client application. To analyze your data, you click on a data item, the numbers on the charts are then filtered by this value, and other data items that are associated with this selected value are then also made available for selection. The screenshot below shows this filtering in action, you can see the selected fields in green, the ones that are greyed out are no longer available for selection.
Transcript
Page 1: Qlikview or Oracle Obiee

QLIKVIEW OR ORACLE OBIEEA Quik Look at QlikView

Outside the world of Oracle, one of the rising stars of the BI world has been Qliktech, the Swedish software company behind QlikView. QlikView is one of a new breed of desktop, in-memory BI tools (see also the forthcoming Gemini tool from Microsoft, an in-memory, column-store BI tool that’ll be delivered as part of Microsoft Office 2010) that are being evaluated by customers as alternatives to traditional relational query tools such as Oracle BI Enterprise Edition and OLAP servers such as Essbase or Oracle OLAP. So what is QlikView, and how does it compare to Oracle’s BI tools?

QlikView, compared to Oracle’s relational and OLAP query tools, has a slightly different approach to storing and querying data. Whilst it can load data from relational and file-based data sources, it then stores the data in-memory in a compressed form with associations defined between data items rather than the traditional joins that we see in relational query tools. The idea behind these associations is that they are derived by Qlikview during the data load based on matching up columns across tables with the same name, and then when you query the resulting dataset you can analyze your data using these associations.

To take an example, the following QlikView demonstration shows data for a consulting company in a Windows thick client application. To analyze your data, you click on a data item, the numbers on the charts are then filtered by this value, and other data items that are associated with this selected value are then also made available for selection. The screenshot below shows this filtering in action, you can see the selected fields in green, the ones that are greyed out are no longer available for selection.

Page 2: Qlikview or Oracle Obiee

Now obviously this way of selecting data items, and this method of “association”, is fairly alien to traditional Oracle BI users and so I put together a simple example using the Sales History (SH) dataset available on recent Oracle databases. The first step in creating some QlikView reports is to define a new document, which in the end will hold the report layout, the data loading queries and the actual reporting data itself, which gets loaded in to memory when you open the QlikView document. Once you’ve defined the new document, you create a script to load data in from in this case an Oracle database.

This is in fact a similar process to putting load scripts together for Essbase and Oracle Express, with connections being made through ODBC and OLE DB for databases and direct connections being made to file and HTTP data sources. I get the impression that the most regular use case for data loading in QlikView is loading from files, and this would correspond to the target market for this tool which is desktop analysts who want to report on their data separately from the infrastructure around their enterprise BI tools and data warehouse.

Once you’ve brought your data in, you can view it either as a database (source) data model, or as the logical model that QlickView generates when it loads in the data. In fact for the SH schema, these are both the same as the joins in the Oracle schema are turned into the associations that QlikView uses, so what you see looks just like the regular sort of data model you’d see in any relational query tool.

Page 3: Qlikview or Oracle Obiee

The joins (or associations) are defined by QlikView during the import process by matching up columns of the same name – each match creates an association. This means that to create a join, you need to alias column names as they come in (to change ID to CUSTOMERID for example, if this is required to join to another CUSTOMERID), and you need to create aliases to break associations so that the STATUS column from the orders table isn’t linked to the STATUS column from the customer table when this isn’t really appropriate.

Page 4: Qlikview or Oracle Obiee

For an Oracle dataset with well defined primary and foreign keys, the logical table structure that QlikView creates in memory is usually pretty much the same as the Oracle schema, but for files or with data sets with lots of cross-related data, QlikView will create synthetic keys and synthetic tables to record the associations and assist with analysis. Internally QlikView tokenizes and compresses the data as well which meant that, for the Sales History data set that I worked with earlier, the total size of the QlikView document (including the data) was only 4MB and took about 1 minute to load and prepare, whereas the corresponding Essbase data set (as built in this OTN article) took up several gigabytes of space and took over an hour to load and prepare. Impressive stuff.

Loading the normalized data set that I used for this article on OBIEE data modeling produced the following logical QlikView table diagram, which as you can see is a direct reflection of the normalized Oracle data model.

This ability to report against normalized data obviously makes setting up the data model easier compared to OBIEE (notwithstanding the need to rename columns to make and break associations), but it then doesn’t make any distinction between columns that are measures and ones that are dimension attributes, which makes working with data in reports a bit more complicated.

Page 5: Qlikview or Oracle Obiee

So once you’ve got the data in, how do you create a report? And more importantly, how do these reports compare to Essbase or OBIEE reports? The actual report creation process is a bit like working with tools such as Hyperion Interactive Reporting or Web Analysis in that you construct an application made up of crosstabs, graphs, gauges and the like, which are then run either in the thick-client Windows application or online using a variety of thick and thin web clients. If I wanted to create a bar chart that allows users to select the product category, month and state of purchase and then graph product sales broken down by customer satisfaction rating, I’d add three field list boxes to my document and then create a bar graph chart, which would be automatically filtered by whatever I select in the lists.

In the screenshot above, I’ve first selected the “Computers” product category which filters the graph, and highlights the states and months that have sales for this category. If I then selected the CA and NY states, only the months that are associated with sales (quantity) for Computers and CA / NY sales would be highlighted, and so on. As such, the “Qlik” part of Qlikview seems to translate to progressive filters of your data based on the items you have selected, with the filtered-down dataset being made up of the data that will join (or “associate”) with the data items that you have selected.

If you’re looking to produce Essbase-like reports where data is shown in hierarchies, this is a bit trickier. There isn’t really a concept of hierarchies in Qlikview; you can create column groups which group together columns that are related, but I can’t see a simple way to select column members, for example, by their position in a hierarchy which is pretty standard in OLAP tools such as Essbase.

Page 6: Qlikview or Oracle Obiee

Displaying data in hierarchies is a bit simpler, especially as QlikView automatically calculates the relationship between columns and their values so that, for example, if you display product categories and products in a pivot table chart, the product are automatically shown under the categories that they relate to.

What I found trickier (as in I couldn’t do it) was to create a pivot table with regions, for example, across the top of the pivot, or to add page items or other controls. I’ve only used QlikView for a few days and so I may well have missed this feature, but none of the example documents that came with either of these features and therefore I suspect that it’s not something that the product does, for the moment.

Working through the examples that ship with QlikView 9, visualizations and fast access to data seem to be the main strengths of the tool. For example, the Sparkline graph type that features strongly in Stephen Few’s “Information Dashboard Design” book is present and correct in QlikView:

Page 7: Qlikview or Oracle Obiee

As are butterfly financial-style tables, used for displaying P&L reports and other finance data.

One of the examples uses Google Maps to illustrate data, in this case around house for sale in London.

Where QlikView is weaker in my view, compared to any of the Essbase-backed query tools, is around number-intensive and hierarchy-intensive applications. It’s not easy to select dimension members via a hierarchy, the scripting and calculation language that comes with the tool obviously doesn’t support the richness of languages like MDX and Essbase calculations scripts, and of course there are not the applications such as Planning and Financial Management that you’d expect to make use of in the Finance Department.

Page 8: Qlikview or Oracle Obiee

That said, for what it does (provide fast access to star and normalized data sets, including data sourced from files and from the web) it does the job very well, with excellent load and preparation times and sub-second response times, even with large data sets. You probably wouldn’t use Qlikview instead of Essbase or Oracle BI EE as they address different requirements, but as a sales analysis tool or a tool to analyze OLTP datasets, it takes some beating.

It wouldn’t surprise me to find that Oracle or another of the big vendors acquires QlikTech at some point as QlikView would make a nice complement to the more enterprise, heavy-duty query tools, especially when tools such as this and the forthcoming Gemini from Microsoft threaten to create desktop BI tool “bridgeheads” on users’ desktops as they are just so easy to use and to set up (can you imagine getting Business Objects and all its technology stack up and running over a couple of evenings, for example, let alone learning how to use it?). It wouldn’t also surprise me if vendors such as Oracle do more to bring their database and query tool processing in-memory, given the amounts of RAM available on users’ desktops these days and some of the advances being made in data compression technologies.

Page 9: Qlikview or Oracle Obiee

OBIEE - Oracle Business Intelligence Suite  Enterprise Edition Plus

Oracle Business Intelligence Suite Enterprise Edition Plus (Oracle BI EE Plus) is a comprehensive suite of enterprise BI products that delivers a full range of analysis and reporting capabilities. Featuring a unified, highly scalable, modern architecture, Oracle BI EE Plus provides intelligence and analytics from data spanning enterprise sources and applications—empowering the largest communities with complete and relevant insight. Oracle BI EE Plus also bundles key Oracle Hyperion reporting products for integrated reporting with Oracle Hyperion financial applications.

From TITAN-Pinnacle you can expect great innovations.

Each TITAN-Pinnacle consultant is expertly trained and certified on Oracle BI EE. As an Oracle Hyperion Preferred Select Partner (one of 8 out of over 300 partners), TITAN-Pinnacle is your choice for the greatest value for your Oracle BI EE consulting dollar. Benefits of Oracle Business Intelligence Enterprise Edition

Comprehensive BI functionality built on a unified infrastructure—Includes interactive dashboards, full ad hoc queries and analysis, proactive intelligence delivery and alerts, enterprise and financial reporting, online analytical processing (OLAP) analysis and presentation, high-volume production reporting, real-time predictive intelligence, disconnected analytics, as well as integration with Microsoft Office.

Pervasive insight to everyone, everywhere—Provides relevant insight to everyone, not just analysts. All levels of the organization can see information optimized for their role.

Unified business model—Built on an enterprise information model that unifies metadata across the Oracle BI tools and analytical applications for the lowest TCO.

Hot-Pluggable BI infrastructure—Integrates with any data source, extraction, transformation and load (ETL) tool, major business application, application server, security infrastructure, portal technology, other front-end and analytical tools, and databases including IBM DB/2, Teradata, Microsoft SQL Server; SAP Business Information Warehouse (BW), Microsoft Analysis Services; flat files; XML data; and unstructured data.

Customers have much more flexibility in change management by allowing them to leverage on what is already built as well as the ability to track compliance an important feature for those companies concerned with Sarbanes-Oxley

Oracle Hyperion can deliver this sort of breakthrough performance management because we offer the most comprehensive set of features, functionality and flexibility in one system Oracle Business Intelligence Enterprise Edition

With Oracle Business Intelligence Enterprise Edition, you can "start anywhere" and drive breakthrough performance everywhere in your enterprise.

Page 10: Qlikview or Oracle Obiee

Quite interesting and timely article as Qlikview has been mentioned very frequently at my current customer site Qlikview has been mentioned and positioned as an alternative for OBIEE for a large upcoming project involving a considerable number of OLTP applications as source on different formats, mostly file based, for operational and desktop analysis reporting. Considering the kind and purpose of the desired reporting, fast access to OLTP datasets and response time to facilitate fast reaction to a reported situation, I do see Qlikview as a good solution, not as a substitute of OBIEE, but as the right tool for the right purpose. Another arguments that has been used for acquiring Qlikview as reporting tool is the pricing compared to OBIEE licensing. Should not be the nr 1 reason for making a decision but it does counts in time of crisis!There one things I was wondering while reading your article; would you still have advantage by creating an integration/consolidation layer for the source data instead of leave it to Qlikview? Once Qlikview bases joins (associations) on columns with the same name it could be useful to have a data layer where naming conventions are applied and eventual data quality conditions and business rules are previously used.

Oracle OBIEE and QlikView are completely different types of solutions, so it will really depend on what your requirements are. Also, I see that you work for a large financial institution, so you probably have lots and lots of data to share. In general, QlikView is more affordable, but only for smaller data sets. As the data volume grows, QlikView doesn't scale easily, and becomes as expensive as Oracle. So there might not be a cost savings there. In that case, you're just looking at features. Oracle has more breadth and depth of functionality, but it takes time and expense to roll them out - it requires greater IT involvement. QlikView is up and running faster, but is less flexible. And as I mentioned, if you're scaling to lots of users, it can become complicated, too.

I haven’t used QlikView enough to know whether it would benefit from denormalized (star schema) tables, as opposed to normalized ones, but I suspect from what I have seen that it doesn’t really matter, although simplified joins declared using FK and PK constraints will make the detection of associations easier and simpler.I would agree with your comment though about an integration layer, which would achieve as you say (a) consistent naming standards, to make and break associations accordingly, and (b) to deal with data quality issues.One thing I would say having worked with it briefly, is that the style of UI and the method of selecting and filtering data might not suit all users and all reporting scenarios. It’s tricky to work with and display hierarchies, and building a new report is more like extending an application rather than creating a report using the kind of report writer you get with Answers or Discoverer. That said, where the tool is strong is on fast access to data (including lots of data), the ease with which you can bring in new data (though I don’t think the integration features are as strong as Oracle BI EE’s federated query feature) and the ease in adding filtering etc.

Page 11: Qlikview or Oracle Obiee

The advantage of using QV is, if we export a report which has more that 65 536 rows into xls sheet it will be exported to sheet 1, sheet2. The Disadvantage is if we try to load lots of data, some time due to system memory and CPU usage the report runs slow even i get "out of virtual memory".These problem is faced by me in QV v8.5.I n OBIEE for analysis (analysis studio). The problem in QV while loading the data is not observed in obiee, since they have restricition to view only 9999 rows of data's. Others can be viewed after exporting into file.

We've got both presentations (IBM's Cognos and QlikTech's QlikView) firstly and decided to purchase QlikView for that main considerent that you've described above.More man power, resources and money spent on training, for Cognos implementation.But, i remember, from the Cognos presentation, one nasty thing: if you make a report and want the user to "Drill-Down", you must build again the same report, but at the drilled-down level and connect them. Something like that. Disgraceful, if not stupid.As for QlikView, it has one major shortcoming (from my point of view): you can't make a select (load) from more than 2 tables residing in memory at some point in time. That's making loading data into QlikView a step-by-step process and that's time consuming. 

Page 12: Qlikview or Oracle Obiee

Qlikview Vs Others

I have worked on Qlikview version 7 and 8 for one year and now presently working on Oracle BIEE(Siebel Analytics) and Hyperion Essbase system 9. Have done dozens of POC and implementations and Qlikview has gained the same respect everywhere. Personally I just love this tool.

I have mentioned some of the important areas where Qlikview is lacking and you will notice that in the coming releases, Qliktech will try to touch all these areas showing some improvements.

I would like to share some pros and cons:1. The biggest disadvantage is the amount or size of data. If you have data huge data in hundreds of GB, then qlikview may not be a good choice.Qlikview was having alimit of 2 Billion rows in a table to be loaded before the release of version 9 but now in version 9 it is not there and the size of data directly depends on system RAM. But even now I did not see a guide telling the RAM requirements for a specified amount of data and how does it increase as the amount of data size grows.still there are some workarounds in certain cases. Design OLAP cubes or Data Marts in your database and keep the logic simple in Qlikview. Rather than creating a huge application, divide into smaller applications and use navigation. But if a single Datamart is huge, Qlikview is not for you.

2. Now speaking for small data sizes, I have observed that it is not the not the pill which cures a patient but it is the diagonosis of a doctor. Dont rapidly start making applications. They may be ready in few days but to manage them could be very difficult and time consuming. plan your needs, take some time to think on it and discuss. when plan is clear in your mind, start working on it.3. For organisations having data growing rapidly, give a thought to create a datawarehouse with marts designed as per your needs. this will have two advantages. Your qlikview application would be simple and if tommarrow you plan to implement some other BI tool, this will be of great help.4. few days back, I just heard about Qlikview providing alert capability. i am not sure if alerts can be sent to hand held devices and cell phones. Also check if these alerts can be designed by business users using some wizard or interface. If alerts can be created using API programming, it will really not be useful. I have seen a wizard based alerts available in Qlikview to define an event and sending the content via email.

5. Proprietary files -Another major disadvantage with all the tools storing data in their proprietary files like Qlikview, Essbase and Cognos. Tomorrow if you plan to have another tool for some reporting or if you plan to have another application talk to these cubes(Qliktech may not like to refer its files as cubes which I beleive is nothing else), then there could be a seroius problem. Major vendors generally provides connectors for major Multidimensional sources like Essbase and Cognos but for tools like Qlikview, it Could be difficult.

some of the features where Qlikview lacks are:1. Alerts- Capability to create alerts and delivers it to not only Email but blackberries, hand held

Page 13: Qlikview or Oracle Obiee

devices, mobile phones etc. In version 9, I heard something for this has come. Also Qlikview do not have any integration with BPEL.2. Multi user development environment- This feature allows multiple developers work on a single project and the utility synchronises the peices of project each developer is working with the main project.Qlikview completely lacks this feature and until version 9, I did not hear anything about this feature coming up.3. Connect and extract data from multi dimentional objects- I guess for SAP BW, the connector is available but not for Hyperion Essbase, Cognos Cubes and Microsoft cubes.4. Export data or metadata to XML- The reports cannot be exported to XML format which is one of the major disadvantage. We can use the XML reports as a data source with another softwares which understands XML.5. XBRL- Qlikview does not support XBRL. Those who do not understand what is XBRL and why it is required and useful, please read this http://www.xbrl.org/WhatIsXBRL/6. Seperate component for operational reporting or production reporting or Popularly known as Pixel perfect reporting (for example Printing Salary Cheques) and report bursting like BI publisher(previously XML publisher) or SQR reporting in Hyperion Essbase. But we can argue here in the sense that Qlikview is an analytical tool and is not made for such needs but definately it restricts you somewhere and do not provide you with an option.7. Integration with Microsoft Office tools- Qlikview just exports the data into an excel file or exports a report object to a png file or using OCX you can do something to make it work with MS office tools(I never tried this). It nowhere stands in comparison with Hyperion Add-In or Cognos Add-In. This is one area which is very important from business users point of view and developing this feature may not be that easy for Qliktech.8. Support for advance features like embedded browser(available in hyperion Interactive reporting), flickers(rolling messeges) etc as an standard options.9. Metadata Management- There is no concept of Physical model, Business Model and Presentation Model. There are no subject areas. The End users had access to all the tables loaded to the Qlikview and the matter becomes difficult when you have large number of tables involved. For end users to create a report themselves, they need to identify the tables and columns from this long list which can make things difficult for them. The solution to this could be to create a seperate Qlikview document for every subject area. This greatly effects managebility and reusability. Those who have worked with Oracle BIEE BI Administrator or Cognos Framework manager or Business Objects Universe will understand the importance of subject areas or Packages.10. Impact Analysis- If you want to make a change in the Qlikview script or datasource and want to know the impact of this change on report objects beforehand, you could be in problem and may have to check each and every report manually and document it.11. Alternate Hirarchies- No concept of Alternate Hirarchies.12. SQL Generation- If you want to see the SQL generated behind a query, you can not see that in Qlikview whereas other tools like OBIEE, Cognos, BO and Essbase allows you to see the SQL/MDX generated for a query. This feature may not be relevant to business users but for developers this helps a lot in their day to day activities like debugging or data validation.13. Resolving circular loops and traps: The only way to resolve circular loopes in Qlikview is to create an alias table wheare as in BO and Cognos you can define contexts which allows you to resolve the circular loops without creating any alias table. When it come to resolving traps like Chasm and Fan traps, Qlikview is helpless.

Page 14: Qlikview or Oracle Obiee

some more issues -1. Security- Qlikview does not have a graphical interface to create roles, users and privilages assigning to roles. Also can users be allowed to log in on certain weekdays only. Also can users change their passwords themselves using some interface. Does the password expires automatically after certain no of days and user will be asked to change it.2. I am not sure if Qlikview can use user directory for external authentication like LDAP database or NTLM. Also can it be used with existing SSO(Single Sign On).3. Does Qlikview has feature to automatically end the session if it is not being used to avoid any session hacking.4. Is there any graphical interface to monitor the sessions in use. Check the queries they have fired, previous requests, kill any current requests, monitor time etc.

There are some good features in Qlikview as well:1. The ability to consolidate the data coming from various sources(databases and flat files) which is in a common format in the data cloud(Qliktech uses this term). this is fantastic. In other OLAP tools sometimes it becomes difficult to join and RDBMS and a flat file data.

2. Many people criticise Qlikview for its associative logic which says the feild names with same names are joined automatically but the designer has the liberty to remove or edit those joins. Even Siebel analytics makes the joins automatically in BI Administrator and Hyperion Essbase sensing same field names.It helps the designer to see the qualified members for joins and make amendments after verifying them


Recommended