Difference Between Direct-Connected Data and Extracted Data

6.0

/

6.0.2

Unified some calculation logics of direct-connected data and extracted data. For details, see section "Calculation Logic of Extracted Data and Direct-Connected Data in Components."

Extracted Data

If you use extracted data, data in the database will be extracted to FineBI (similar to saving data to FineBI). Therefore, data in the database and data in FineBI are not continuously synchronous. You need to regularly update data in FineBI so that data can be consistent with that in the database. Since data is extracted and saved to the FineBI engine, enough space in the local disk is required in the Extracted Data mode.

Through data cache/data copy creation, online analytical processing (OLAP) of large volumes of data is supported to accelerate query performance, ensure analysis experience, and minimize the impacts on business databases as much as possible.

Direct-Connected   Data

If you use a direct -connected dataset, data in your database is directly used in FineBI for calculation. Therefore, data in FineBl and data in the database are kept synchronous.

With the help of your big data platform/data warehouse, the simple self-service analysis requirement can be met in the case of high concurrency and large data volumes.

Extracted Data

For details about related server performance requirements for extraction, see Recommended Environment and Configuration for Project Deployment.

Direct-Connected   Data

• The self-owned big data platform boasts high database reliability and excellent performance.

• Self-service analysis is performed according to the usage modes recommended for different database types respectively.

•  Web clusters can be used to connect to databases to support high concurrency.

• Certain database O&M capabilities are guaranteed.

Extracted Data

(1) Complex self-service lineage analysis of 10 million data records

•  10 million data records (10 million or less)

•  Complex dataset

• Ultra-fast analysis (what you see is what you get)

(2) Joint analysis of data from multiple databases

Direct-Connected Data

(1) Simple self-service analysis of large volumes of data

• Simple query of over 100 million data records (generally self-service datasets not recommended)

• SQL query in lieu of visual operation

(2) Relatively-high user quantity and concurrency, requiring linear scalability

(3) Relatively-high requirement on timeliness

(4) Relatively-high requirement on data security

(5) Scenario where the data volume is not large and extraction is troublesome

Extracted Data

Due to the data volume limitation, extraction of over 100 million data records cannot be supported.
The data needs to be fully extracted from the data warehouse and business database to the FineBI server. Some customers may not want to fully extract the data again.

Direct-Connected   Data

• For more than 100 million data records, the performance relies entirely on customers' databases and data warehouses. BI cannot perform deep optimization. (Extraction of data from external databases, which are often not BI-dedicated databases, for query will occupy database resources and cause poor performance.)

• The timeliness of the data depends on that of the data in customers' warehouses.

• The multiple levels of self-service datasets result in complex SQL query and longer calculation time.

• In the Extracted Data mode, joint analysis (Add Association, Join, Union All, Column from Other Tables, and Model View > Edit Association) cannot be performed on data tables across different data sources.

Extracted Data

Generally, it is applicable to small enterprises that have small data volumes/low budgets and require self-service analysis.

Direct-Connected Data

Generally, it is applicable to large enterprises, who have completely-constructed big data platforms and attach great importance to data security (indicating that they do not want to extract data again) and data timeliness.

Influence of the filtering and quick calculation on the summary value

No influence

No influence

Influence of one indicator (for which the filtering and quick calculation are performed) on other   indicators (for which the quick calculation is performed)

No influence

No influence

Influence of one indicator (for which the filtering and quick calculation are performed) on other   indicators summarized by the quick calculation

No influence

No influence

Indicator-based dimension filtering/sorting

Filtering/Sorting based on the results of summary rows

Filtering/Sorting based on the results of summary rows

Filtering logic of the cross table

Normal filtering according to the filtering conditions

Normal filtering according to the filtering conditions

Filtering in Filter and hear filtering at the same filtering level

Intersection of the filtering conditions in both

Intersection of the filtering conditions in both

Different filtering logics for null and empty strings

Filtering out all null and empty strings (whether you select null or empty strings for filtering)

Filtering according to the database logic

(If the database logic is to filter out null and empty strings separately, the result of direct-connected data will be different from that of extracted data.)