Evgeny Faizrakhmanov

National Research Tomsk Polytechnic University, Russian Federation

OLAP classification by means of data stores organization. Advantages and disadvantages

Currently most of the enterprises use databases as corporate information storage.  As a consequence, this common usage of the databases at the diverse levels of management has resulted in the bulk data accumulation. The data is expected to be used for the enterprise statistical model and development strategy analysis and building, thus to ease the executive decision-making process. For this purpose the data have to be introduced in appropriate and convenient form for analyst. That is where we can employ OLAP technologies.

        The OLAP concept basis is multidimensional data presentation. However, in the dozen of publications the abbreviation OLAP means not only multidimensional conception, but the data storage in multidimensional database.   E. F. Codd, who has defined OLAP requirements, points out the following ‘Relational DB is the most approachable technology for the corporate data storage.   The necessity appears not in the invention of the new DB technology, but in the analysis tools completing existing DBMS that are  flexible enough to  predict and automatize different types of intellectual analysis specific for OLAP’          Nowadays exist  three basic OLAP models differentiated by the database organization method: 

·        MOLAP, Multidimensional OLAP;

·        ROLAP, Relational OLAP;

·        HOLAP, Hybrid OLAP.

Multidimensional OLAP.

        As for multidimensional OLAP systems cube structure is stored in the multidimensional database together with the preprocessed aggregates and sheet levels copies.  Therefore all data requests are complied with multidimensional database system what makes MOLAP systems exceedingly fast-acting. MOLAP design demands bigger disc space span due to the storage in multidimensional database of sheet levels copies.  Nevertheless excess space span is not very large because the data in MOLAP is effectively stored.

        The advantages of MOLAP are:

·        The whole data is store in multidimensional structure, what essentially increases the speed of data processing;

·        The expanded libraries are available for the composite functions of the operational analysis;

·        The sparse data processing is conducted much faster than in ROLAP systems.

The disadvantages are:

·        The necessity of the dedicated tooling for cubes formation and there conversion in case of the basic values changes;

·        Difficult measure changing without any reaggregation.

        Relational OLAP.

As for multidimensional OLAP systems cube structure is stored in the relational database. The lowest level measures are stored in relational data mart used as data source for cube. Pretreated aggregates are also stored in a relational table. Because of the  relational tables ROLAP architecture allows you to store large amounts of data. ROLAP-systems do not lag in terms of sheet data as ROLAP architecture sheet values ​​are taken directly from the data marts.  

The advantages of ROLAP- systems are:

·        The ability to use ROLAP data warehousing and OLTP- different systems;

·        The possibility of manipulating large amounts of data ;

·        Security and administration provided RDBMS .

Disadvantages:

·        Obtaining of the aggregates and data sheets is conducted slower than in MOLAP and HOLAP;

·        Limited by the functionality of the SQL systems as analytical user requests are translated into SQL-operators hauling;

·        Difficult to count the aggregated value with changes in the initial data;

·        Difficult to maintain aggregate tables.

Hybrid OLAP.

Hybrid OLAP combines features of MOLAP and ROLAP. HOLAP models use advantages of both architectures  and minimize the disadvantages. In HOLAP-systems the cube structure and pretreated aggregations are stored in multidimensional database. This allows quick extraction of aggregates from  MOLAP structures. Values ​​of the hierarchy lower level  in HOLAP remain in the relational data mart, which serves as a data source for the cube.

The advantages of the approach are the combination of ROLAP technology for the sparse data and MOLAP for dense areas, and the disadvantage is the need to maintain the MOLAP and ROLAP.

It can be concluded that the choice of the model depends on OLAP requirements on data download rate, the required disk space, etc. After you select and configure OLAP model corporate enterprise information will be presented to the analyst in the required form, which allows reviewing the strategy for further development of the company.

 

The list of references:

1.     MOLAP, ROLAP AND HOLAP. – URL: http://www.1keydata.com/datawarehousing/molap-rolap.html

2.     Types of OLAP Systems. – URL: http://olap.com/types-of-olap-systems/

3.     Dimensional Relational vs. OLAP: The Final Deployment Conundrum. –URL: http://www.informationweek.com/software/information-management/dimensional-relational-vs-olap-the-final-deployment-conundrum/d/d-id/1054577?