Saturday, May 28, 2011

Qlikview is now a Leader


As I was expecting, Qlikview joined the leaders quadrant in Gartner's magic quadrant for BI 2011.
Qlikview is cited as a self contained BI Platform and the strengths being interactive, great visualization and end user friendliness and satisfaction. I am very happy about it.
But I am more focused on seeing the challenges ahead. It will be interesting to see how Qlikview maintain that position and stand in the competition.
The challenges cited by Gartner are
1. Lack of expansive product strategy
2. Limited metadata management
3. Lack of broad (high volume) BI deployments
4. Lack of Microsoft office integration
5. Poor Performance when data volume and number of Users increases.

The findings are not new and Qliktech surely needs to seriously think about these shortcomings.
I want to discuss further on the above points in detail.

1. Lack of expansive product strategy : To compete with large vendors like Oracle, it becomes very important to have a competent product expansion strategy. Oracle has very aggressive product strategy and has a vision to integrate its various offerings like Oracle BI, Hyperion Essbase, Oracle Enterprise performance management and more importantly their pre built analytic models popularly known as BI Applications. Though Qliktech has already taken one step in this direction by targeting application vendors like Salesforce and can offer pre built models for Salesforce customers but this is not enough. Qliktech has to work agressively in developing such pre built models for other but big applications. EPM is one area which is still untouched and lack of vision in this area can be disastrous and will simply throw Qliktech out of competition. Vendors not only should now think of Softwares but also start thinking about offering Hardware configured for optimum and enhanced performance. Oracle has got its popular Oracle Exadata, its database pre configured with HP's hardware and is agressively promoting it.

2. Limited metadata management : Qlikview offers limited metadata management capabilities and the primary reason I see is because Qlikview is focussed on small scale or much smaller than average size deployments, it did not see much relevance of metadata management. This can be dangerous to them as well as their clients as when they grow, they will start seeing the need for it and would require the investment they tried to save at the beginning. Even if Qliktech decides to go for building its capabilities in metadata management, the basic problem for them will be to start believing in OLAP dimensional building which will be against their basic principles. Qliktech market its product as a non OLAP tool which actually is not and treat the underlying data as a cloud in the memory. Hence when it will see the need for conforming dimensions to do cross functional analysis, it may become a matter of choice rather than a matter of capability.

3. Lack of broad (high volume) BI deployments: For Qliktech as mentioned above and as cited in Gartner's report, the major challenge will be to deploy large scale applications. As of now they have proved their capabilities in small or much smaller than average scale deployments and I think that is what Qlikview was made for. One of the Qliktech's selling point is that Qlikview do not require a datawarehouse. Now this same selling point will stop them to move ahead or prove their capabilities in average and large scale deployments.
For those who want to know why, please read one of my earlier post here
This again will depend on reviewing its sales strategy and making corrections to their basic beliefs which is not going to be an easy task. If they do not start using the terms datawarehouse and OLAP, it will difficult to maintain the Leaders position.

4. Lack of Microsoft office integration: This is something I have mentioned in one of my post in Year 2008 read here. It seems Qliktech is least bothered. Its current capabilities are very basic in terms of simple export to MS Excel. In coming releases if it do not develop such capabilities, it will he hard for Gartner or Forrester to give a space to Qliktech in their reports and compare Qlikview with Oracle or IBM. There are many more such features which I have mentioned in my post earlier. Some of them which are important according to me are building connectors for their proprietary QVD and QVW files so that their models can be available to other applications, SQL generation queries to help developers in debugging etc.

5. Poor Performance when data volume and number of Users increases: This is again linked to point number 3 above.

Feel free to post your comments or thoughts.

Till next time

Manohar Rana

Saturday, April 16, 2011

Enhance Business Intelligence Performance

Hi All,

In any Business Intelligence Implementation, the key factor is the performance. Performance factor always plays a key role in User accepting or liking the application.
We should do everything possible to enhance the performance and here are some tips some of which are very generic and can be used in any BI Implementation.
From a solution perspective:
1. Use of Datawarehouse: Though a datawarehouse is not compulsory for any BI Implementation, we cannot simply think about a BI solution without a datawarehouse because of the advantages it offers in terms of performance and simplicity. This is important for small implementations who sometimes neglect and underestimate the use of datawarehouse.
From a BI Tool Perspective:
1. For every tool it is important to reduce the size of the application by removing the unnecessary objects.
2. Try to create different database connections for different set of users based on the priority.
3. Try to create a seperate database connection for populating any session level variables.
4. Try to make the best use of system Cache. If the tool allows to cache the results of the queries, use it and if possible pre populate the queries which are very frequent and used mostly.
5. Minimise the calculations happening at the BI level by pre calculating them in datawarehouse.
From a database perspective:
1. The most important thing is to perform every possible calculation you can do in database. We very frequently neglect this saying this is a small thing and cal be calculated or performed at BI level. We should avoid this and if something is possible in ETL or database, do it here even it cost you adding a few extra columns or tables.
2. If you can create a perfect star models, nothing like that.
3. Try to use the database techniques like Partitioning and indexing to enhance the performance of database queries.

There may be several other tips and techniques which we can follow to improve performance and if you have any, please feel free to share.
Till next time.

Manohar Rana